Methods
Ethics statement
The study protocol was approved by the Institutional Review Board of Korea University Anam Hospital (IRB No. ED11094). Written informed consents were obtained from all patients.
Patients and study design
Patients who planned to undergo elective hepatobiliary surgery in Korea University Anam Hospital were sequentially enrolled from July 2011 to January 2013. Patients <18 years of age, with baseline estimated glomerular filtration rate (eGFR) of <60 ml/min/1.73 m2, on maintenance renal replacement therapy (RRT), or who developed AKI preoperatively were excluded. Blood and urine samples were obtained before surgery (serum NGAL-Pre and urine NGAL-Pre). Urine samples were also obtained postoperatively at 6, 12, and 24 h (urine NGAL-6 h, NGAL-12 h, NGAL-24 h), and serum samples at 24 h (serum NGAL-24 h). Patients were followed up at the outpatient clinic after discharge, and serum creatinine levels were measured at 1, 3, and 6 months after the surgery.
Postoperative clinical AKI was diagnosed according to the Acute Kidney Injury Network (AKIN) criteria (i.e., ≥0.3 mg/dL or a ≥50% increase in the serum creatinine level from the baseline value within 48 h, or a urine output <0.5 mL/kg/hr for ≥6 h). Subclinical AKI was defined as an increase in the serum or urine biomarker (NGAL in this study) level above the determined cutoff value, without meeting the AKIN criteria [
2‐
4]. This definition was also used in a multicenter pooled analysis by Hasse et al. and patients with subclinical AKI showed poorer outcome compared to non-AKI group who had no increase of both NGAL and serum creatinine [
2]. The eGFR was estimated using the Modification of Diet in Renal Disease equation.
We first assessed the diagnostic ability of NGAL for the AKIN criteria-based clinical AKI. Then, we divided the patients into two groups: those with AKI (clinical or subclinical) and those without AKI. Baseline characteristics, surgical characteristics, and outcome variables, including death, length of hospital stay, and decrease of eGFR during the follow-up period, were compared between the two groups.
The primary outcome was the development of postoperative AKI, and the secondary outcomes were the length of hospitalization, postoperative death and the decrease in eGFR at 6 months after the surgery.
Laboratory measurements
Collected blood and urine samples were centrifuged, and the supernatants were stored immediately at -80°C until the biomarker assay. Serum and urine NGAL were measured using NGAL ELISA kit (BioPorto, Gentofte, Denmark) according to the manufacturer’s instructions.
Statistical analysis
SPSS software, version 19.0, was used for statistical analyses. Comparisons between the two groups were performed by the Student’s t-test or the Mann–Whitney U test for numerical data, and χ2 test or Fisher’s exact test for categorical data. To examine the diagnostic performance of urine NGAL for the development of clinical AKI, receiver operating characteristic (ROC) analysis was performed. We conducted multivariate logistic regression analysis to identify the risk factors for the development of clinical or subclinical AKI, and multiple linear regression analysis to identify the predictors for the extent of decrease in eGFR at 6 months postsurgery.
Discussion
AKI is an important postoperative complication that increases morbidity and mortality, with an incidence ranging from 10% to 30% after a major surgery [
5,
6]. In a retrospective study of 10,518 patients with normal renal function, postoperative AKI was associated with increased mortality even in patients who recovered their renal function at the time of hospital discharge [
5]. In another study, patients with postoperative AKI were reported to have more dependency in the activities of daily living at 6 months after discharge [
6].
In contrast to cardiac or vascular surgery, the incidence or risk factors of AKI after intraperitoneal surgery are not well known due to the paucity of data. In our study population of patients with a baseline eGFR of ≥60 ml/min/1.73 m
2 who underwent elective hepatobiliary surgery, 7.6% developed AKIN criteria-based clinical AKI but none needed RRT. The overall incidence and severity of AKI were much lower than those in a previous report that demonstrated an incidence of AKI of 15.1% after hepatic resection [
7]. Although direct comparison of the patients’ baseline and operative characteristics is difficult, the higher prevalence of chronic kidney disease (CKD) (12.8%) in the previous study might be one reason for this discrepancy.
We next tested the validity of serum and urine NGAL in the early detection of AKI. NGAL is the most widely studied biomarker for AKI; however, most studies have been performed in patients undergoing cardiac surgery or transplantation [
8‐
13]. As expected, we found that urine NGAL obtained at 12 h postsurgery was the most predictive parameter for the detection of AKI, with a cutoff value of 92.85 ng/mL. Serum NGAL did not show predictive value in our population partly because it can rise in liver injury. Organs other than kidney, such as liver, lung, and gastrointestinal tract, also express NGAL under physiologic condition and the expression is increased under injured or post-operative condition [
14]. Our previous study with critically ill patients and another study with liver transplant patients also presented the usefulness of urinary NGAL compared to plasma NGAL in the prediction of AKI or prognosis [
15,
16]. Although another study performed in patients undergoing cardiac surgery presented a similar value (87 ng/mL) [
11], the optimal cutoff values for detecting postoperative AKI vary across studies. In addition, defining the proper time point for sample collection is also important. Several previous studies performed in patients undergoing cardiac surgery or liver transplantation suggest that NGAL levels at 0–6 h postsurgery are meaningful for the early detection of AKI [
8‐
11,
13,
17], whereas a study performed in patients undergoing non-cardiac major surgery showed no correlation of NGAL levels at 2 or 6 h postsurgery with the development of AKI [
18]. In our study, we also observed that urine NGAL-12 h, but not NGAL-6 h, reliably predicts clinical AKI in patients undergoing hepatobiliary surgeries. Concerning the proper cutoff value or time point, further larger studies with different types of surgeries are needed.
Recent advances in biomarker research have made it possible to identify patients who showed only increased levels of tubular injury markers without fulfilling the consensus criteria for AKI. Those patients could not have been recognized in the pre-biomarker era because their urine output and serum creatinine levels are well maintained. However, several recent studies demonstrated that patients with this type of subclinical AKI comprises 15-20% of patients admitted to an intensive care unit or an emergency department, and they have increased risk of death or need for dialysis compared with patients without elevated levels of tubular damage markers [
2,
19]. In our study, with an NGAL cutoff value of 92.85 ng/mL, >30% of patients who underwent elective hepatobiliary surgery developed subclinical AKI. We then included these patients in the AKI group (n =52, 39.7%) and compared their preoperative risk factors with those of the non-AKI group.
Among several factors that were significantly different between the AKI and non-AKI groups in univariate analysis, older age and higher MELD-Na score were found to be independent preoperative risk factors for AKI after hepatobiliary surgery. Although several models for predicting AKI after cardiac surgery, such as Continuous Improvement in Cardiac Surgery Study and Cleveland Clinic Score, have been proposed and validated extensively [
20], only few studies developed a prediction model for AKI in non-cardiac surgeries [
7,
21]. Kheterpal et al., by reviewing 152,244 operations, identified 11 independent preoperative predictors of AKI after general surgeries: age ≥56 years, male sex, emergency surgery, intraperitoneal surgery, diabetes mellitus necessitating oral or insulin therapy, active congestive heart failure, ascites, hypertension, and mild to moderate preoperative renal insufficiency [
21]. More recently, Slankamenac et al. developed and validated an AKI prediction score system in patients undergoing liver resection, similar with our study. In that study, they noted that postoperative AKI developed in 15.1% and that preoperatively elevated ALT level, preexisting cardiovascular disease, chronic renal failure, and diabetes were the strongest predictors of AKI defined by the RIFLE criteria [
7]. Although our study did not aim to develop a prediction model for postoperative AKI and had a small sample size for validating a new model, we could identify that old age and high MELD-Na score were significant risk factors for postoperative AKI after hepatobiliary surgery. On the other hand, ALT or the presence of diabetes was not associated with the development of postoperative AKI in our study. The major differences between our study and Slankamenac et al.’s were that we excluded patients with CKD or those who underwent emergency surgeries and included patients with subclinical AKI. These results could suggest that in patients without CKD, age and the degree of underlying liver dysfunction might be more important in the development of postoperative AKI.
It is noteworthy that preoperative MELD-Na score that includes the INR of prothrombin time, and bilirubin, serum creatinine, and Na levels can be considered risk factors for postoperative AKI even in patients with normal renal function. No study to date has investigated the predictive role of the scoring system in postoperative complications. The MELD incorporating serum sodium, known as “MELD-Na”, was developed for the outcome prediction of cirrhosis and is considered to have better prognostic ability than the Child-Pugh classification or the MELD [
22,
23]. In addition, the MELD-Na score has also been reported to be significantly related to survival in patients with hepatocellular carcinoma or pancreatic and gastric cancer-related ascites [
24‐
26]. Similarly, we also observed that a high preoperative MELD-Na score was independently associated with the development of AKI after hepatobiliary surgery. We chose the MELD-Na score instead of the MELD score because we observed that the Na level was significantly lower in the AKI group in univariate analysis, in addition to the significantly prolonged prothrombin time (INR), elevated serum bilirubin level, and lower albumin level.
Interestingly, we assigned patients with subclinical AKI to the AKI group and compared the outcomes. While dialysis requiring AKI is a well-known factor of poor outcome of AKI such as longer hospital stay, higher mortality, and predisposition to end-stage renal disease, we observed in our study that patients with stage 1, 2, or 3 without the need for RRT or even those with subclinical AKI had significantly longer hospital stay and worse 6 months eGFR than the non-AKI group. The difference was also confirmed by a separate analysis between patients with subclinical AKI and the non-AKI group patients. To the best of our knowledge, this is the first study that demonstrated that even subclinical postoperative AKI might portend poor long-term renal outcome. Although our patients with subclinical AKI still had an eGFR of 80 ml/min/1.73 m2 after 6 months of observation, further studies with a longer follow-up period would show a larger impact on patients’ renal outcome. Moreover, the maximum level of urine NGAL, which was used for the diagnosis of subclinical AKI, was demonstrated to be independently associated with the extent of loss of renal function at 6 months postsurgery, even after the adjustment for other variables including liver transplantation.
All these findings in our study could strengthen the important role of urine NGAL as a sensitive biomarker and an outcome predictor of postoperative AKI, and also emphasize the need for recognizing and evaluating patients who develop subclinical AKI.
Along with urine NGAL, having liver transplantation was also demonstrated to be independently associated with poor long-term postoperative renal outcome. In addition to the possibility that frequent development of AKI after liver transplantation leads to worse renal outcome, another possibility for worse 6 month renal outcome in liver transplantation might include the unavoidable use of immunosuppressive drugs such as calcineurin inhibitors. Another factor that was independently associated with worse 6 months renal outcome was a positive fluid balance during the operation. Previous reports also highlighted the importance of fluid overload as a risk factor for death and lack of renal recovery in patients with AKI; therefore, restricted perioperative fluid administration was recommended to reduce postoperative complications and mortality [
27‐
33]. Our study, by demonstrating worse long-term renal outcome, reaffirms the negative impact of positive fluid balance in hepatobiliary surgeries. This association was still significant even after the adjustment for liver transplantation, age, or MELD-Na score.
Despite several novel findings, this study has some limitations. It is a single-center prospective study, and its small sample size restricted the development or validation of a model for predicting postoperative AKI or long-term renal outcome. In addition, we only enrolled patients with normal renal function and only four patients had heart diseases, which may limit the general application of the results of this study. We excluded patients with eGFR <60 ml/min/1.73 m2 because decreased eGFR can be a confounding factor in interpreting the incidence of post-operative AKI and the 6 months renal outcome post surgery. Inclusion of patients irrespective of the type of hepatobiliary surgery, including liver transplantation, might also be a limitation. However, we adjusted for liver transplantation when analyzing the risk factors for 6 months renal outcome. Furthermore, as already mentioned, more frequent follow-up examinations of serum or urine NGAL levels may show additional information, although urine NGAL-12 h was the most predictive parameter for determining AKI in our patients. Finally, we observed the patients for only 6 months. Although our study identified several important findings, these results need to be further validated in a larger cohort of patients with a longer follow-up period.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
EC participated in the conception, acquisition, analysis and interpretation of data, and provided intellectual content of critical importance to the work, and approved final version to be submitted. SCK, MGK, WYC, HKK participated in the analysis and interpretation of data, and provided intellectual content of critical importance to the work, and approved final version to be submitted. SKJ participated in the conception, analysis and interpretation of data, and revised the article, and provided intellectual content of critical importance to the work, and approved final version to be submitted. All authors read and approved the final manuscript.