The importance of renal function and its effect on outcome in liver transplant recipients are well described.1–4 Renal function, as estimated by serum creatinine (Cr), is one variable that is used to calculate the Model for End-Stage Liver Disease score and is recognized as an important variable affecting the morbidity and mortality of liver transplant candidates and recipients. Long-term deterioration of renal function in liver recipients is common and often can lead to chronic kidney disease or even end-stage renal disease.5 Furthermore, the development of acute kidney injury (AKI) has been associated with markedly increased costs and consumption of health care resources in the general hospital population6 and liver transplantation population.7 Increasingly, AKI is also considered an important factor for the development of chronic kidney disease.8 Hence, every effort has to be undertaken to preserve renal function throughout all stages of patient care.9, 10
Investigators have only recently studied the contribution of the intraoperative management of liver transplant patients to the development of AKI and its potential impact on outcome.11–13 These studies have identified several risk factors and predictors for perioperative AKI, but most have been limited by the retrospective nature of the studies. Furthermore, all the studies have been based on perioperative Cr changes. The time sequence of these Cr changes frequently becomes apparent only at a time point when renal injury is already fully developed, and medical management is mainly confined to supportive therapy. In order to better ameliorate AKI and design prospective randomized interventional trials, better surrogates of renal injury are necessary. The recent introduction of markers of renal function and injury will likely overcome this shortcoming. Cystatin C has been introduced as a marker of renal function that is superior to Cr because it is not affected by race, muscle mass, or gender. New markers of renal injury include interleukin 18, kidney injury molecule 1, and neutrophil gelatinase-associated lipocalin (NGAL).14 NGAL, a 25-kDA protein belonging to the lipocalin superfamily, was found initially in activated neutrophils and subsequently in many other tissues, including renal tubular cells, in response to a variety of insults. The determination of NGAL has been conducted in urine and plasma samples.15 It has been successfully used in cardiac studies and renal transplantation studies.16, 17 One of the main advantages of measuring serum or urine NGAL is the relatively rapid time course of changes with respect to renal injury (in comparison with serum Cr levels); changes usually occur within several hours after the insult. The primary goal of this prospective observational study was to evaluate serum NGAL as an early marker of renal injury during liver transplantation and determine its association with commonly established AKI criteria, that is, the RIFLE (risk, injury, failure, loss, and end-stage kidney disease) criteria.14, 18
AKI, acute kidney injury; ASA, American Society of Anesthesiologists; BMI, body mass index; Cr, creatinine; CVP, central venous pressure; FFP, fresh frozen plasma; GFR, glomerular filtration rate; HCC, hepatocellular carcinoma; ICU, intensive care unit; NGAL, neutrophil gelatinase-associated lipocalin; NS, not significant; PRBC, packed red blood cell; RIFLE, risk, injury, failure, loss, and end-stage kidney disease.
PATIENTS AND METHODS
After the study was approved by the institutional review boards at the University of California San Francisco (San Francisco, CA) and Vanderbilt University (Nashville, TN), informed consent was obtained from all patients participating in this study. We prospectively enrolled patients with end-stage liver disease who were scheduled to undergo liver transplantation at the University of California San Francisco or Vanderbilt University during the period from October 2007 to October 2008. Patients were excluded if they were less than 18 years old or required a combined liver kidney transplant. All demographic patient characteristics and select intraoperative variables and laboratory values were obtained from the automated anesthesia record system at both institutions. Average central venous pressure calculations were based on 1154 individually recorded values. Intraoperative mean glucose values and average mean blood pressure values during the anhepatic phase were calculated on the basis of 354 and 545 individual values, respectively. Hospital records and hospital computerized databases were then reviewed, and preoperative and postoperative laboratory values were obtained.
Blood samples for the determination of NGAL were drawn at 3 different time points: immediately after the induction of anesthesia (ie, the baseline; NGAL1), 2 hours after reperfusion (NGAL2), and 24 hours after reperfusion (NGAL3).
The anesthetic technique was the same: induction with propofol (1–2 mg/kg), fentanyl (2 μg/kg), and a nonpolarizing muscle relaxant or, in select cases, succinylcholine. Maintenance of anesthesia was obtained with desflurane or isoflurane (the minimal alveolar concentration was between 0.6 and 1.0). Fentanyl was given either as a bolus or as a continuous infusion as determined by the attending anesthesiologist. Patients were ventilated with a fraction of inspired oxygen of 0.5 to 1, tidal volumes of 7 to 10 mL/kg, and a positive end expiratory pressure of 0 to 5 cm H2O. Transfusion triggers were similar, with a target hematocrit of 25 to 28 and a platelet count greater than 50,000/μL. Fresh frozen plasma was administered at the discretion of the attending anesthesiologist and according to whether thromboelastography was used or not.19 All patients received intraoperative methylprednisolone (1000 mg at the University of California San Francisco and 500–750 mg at Vanderbilt).
All donor organ allografts were implanted either by caval replacement with bicaval anastomoses or by piggyback techniques. Venovenous bypass was used in 1 patient at Vanderbilt University.
Clinical Evaluation of Renal Function and Injury
Serum Cr levels were determined preoperatively and then daily up to 5 days after surgery. Serum Cr levels were used as a surrogate for renal function. However, the term renal function, when used clinically, is often taken to reflect the glomerular filtration rate (GFR) or the nearest approximation of GFR. Hence, we calculated GFR with the Modification of Diet in Renal Disease Study equation (http://www.nephron.com/MDRD_GFR.cgi) on the basis of obtained serum Cr levels and patient demographics.
The RIFLE criteria were used to define AKI. The RIFLE categories can be defined by either GFR determined with serum Cr or urine output. In this study, GFR based on serum Cr was used because accurate urine output was not available at all time points. Risk is defined as a serum Cr elevation of at least 50%. Injury is considered when the serum Cr level increases by 100%, and failure is considered when the Cr level is 300% higher than the baseline.10, 18
Determination of NGAL
Blood samples were drawn at the predetermined time points and processed within 2 hours after collection. Blood collected in serum separator tubes was allowed to clot for 15 to 20 minutes and then was centrifuged for 12 minutes at 1000g. Serum was collected and subsequently frozen at −20°Celsius until further analysis.
All samples were analyzed in batches in a random fashion. Control samples were obtained from 5 healthy subjects (3 females and 2 males). Quantitative NGAL levels were measured with a sandwich enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, MN) according to the manufacturer's instructions. Concentrated samples were diluted up to 80-fold with the manufacturer-provided diluent. Standards, samples, and controls were run in duplicate, and the resulting chromagen was read at 450 nm with an additional 570-nm wavelength correction (Tecan, San Jose, CA). NGAL concentrations (ng/mL) were then calculated on the basis of the constructed standard curves on respective enzyme-linked immunosorbent assay plates. Standard curve R2 values ranged from 0.9903 to 0.9987.
Categorical variables were expressed as number (%), and continuous variables were expressed as means and standard deviations.
The outcome, AKI, was defined as an increase of Cr from the baseline of 50% or greater within the first 2 postoperative days. For the univariate analysis of risk factors for AKI, continuous variables were analyzed by logistic regression. Categorical variables were analyzed with the chi-square test. Differences between institutions were analyzed with the Mann-Whitney U test for continuous variables and with the chi-square test for categorical variables. Correlations between continuous variables were analyzed by the Spearman rank correlation.
Because of substantial interactions between several variables, more than 1 possible multivariate model could be developed for all patients included or for patients with Cr less than 1.5 mg/dL. The final multivariate logistic regression model included potential risk factors that showed significant effects in the univariate logistic model, and measurements with biological or clinical importance were included in the multivariate model as covariates. After highly significant variables were entered into the model, all variables were tested for entry into the model, whether they were significant by univariate analysis or not.
All statistical analyses were performed with JMP 7.0 software (SAS Institute, Inc., Cary, NC). P values less than 0.05 were considered to be statistically significant.
A total of 59 patients were enrolled in this study. The recipient age was 54 ± 9 years, with a gender distribution of 40 males and 19 females. The mean body mass index was 29.1 ± 6.2, and the average Model for End-Stage Liver Disease score was 21 ± 10, which was unadjusted for hepatocellular carcinoma (n = 24, 41%).The complete demographics of all enrolled patients are shown in Table 1.
Table 1. Demographics
NOTE: Data are presented as the mean and standard deviation or number (%).
Abbreviations: ASA, American Society of Anesthesiologists; BMI, body mass index; HCC, hepatocellular carcinoma; MELD, Model for End-Stage Liver Disease.
54 ± 9
40 (68%)/19 (32%)
29.1 ± 6.2
21 ± 10
31 (53%)/28 (47%)
Table 2 depicts perioperative serum Cr levels, estimated GFRs, respective NGAL concentrations, and perioperative variables of all recipients. A subset of patients (n = 14) experienced preoperative renal dysfunction or hepatorenal syndrome (Cr of 1.5 mg/dL or greater). These patients can be identified a priori as being at high risk for further renal injury, and their injury response (Cr and NGAL concentrations/pharmacokinetics) is likely to be different than that of patients with normal renal function.20 Of note is the elevated baseline NGAL concentration in all subjects in comparison with subjects with a Cr level of less than 1.5 mg/dL (Tables 2 and 3). Because the focus of this study was the identification of de novo AKI in patients with preoperatively preserved renal function, we subsequently analyzed patients with Cr concentrations less than 1.5 mg/dL. A detailed analysis of the NGAL pattern and development of AKI was performed in 45 patients (Table 3). Table 3 presents the univariate analysis for variables that were associated with patients at risk for AKI (a 50% or greater increase in Cr in comparison with baseline Cr). More than 50% of the examined patients (24 out of 45) fulfilled this criterion.
Table 2. Univariate Analysis Including All Baseline Cr Values
Acute Kidney Injury (>50% Increase in Cr)
NOTE: Pressors include a combination of the phenylephrine dose and 5 times the norepinephrine dose.
Specifically, we were interested in whether the baseline NGAL or NGAL drawn approximately 2 hours after reperfusion could identify patients at risk for AKI. Very early detection would allow better design of interventional studies and ameliorate evolving renal injury.
Figure 1A demonstrates all NGAL2 and Cr concentrations during the first 48 hours in all patients. Figure 1B depicts NGAL2 and Cr concentrations in patients with a Cr concentration less than 1.5 mg/dL.
When all patients were included, absolute serum NGAL2 concentrations were not a great predictor of AKI, but a delta NGAL (NGAL2 − NGAL1) achieved a reasonable model (Table 4).
Table 4. Multivariate Analysis
All Cr Values
Baseline Cr < 1.5 mg/dL
NOTE: NGAL1 refers to baseline measurements; NGAL2 refers to measurements taken 2 hours after the anhepatic period.
Abbreviations: Cr, creatinine; NGAL, neutrophil gelatinase-associated lipocalin; NS, not significant.
NGAL2 − NGAL1
In patients with a Cr concentration of less than 1.5 mg/dL, NGAL was the most significant variable in all cases in predicting AKI (Table 4). A calculated area under the receiver operating characteristic curve of 0.79 and a cutoff for NGAL of 139 ng/mL, providing the best sensitivity and specificity, are reasonable in view of the relatively small cohort size (Fig. 2).
There was a substantial correlation between the institution, piggyback technique, and baseline GFR. The type of surgical technique was associated with the development of AKI. The use of full caval clamping was significantly associated with an increased risk for AKI (Table 4). There was a trend of lower mean blood pressures during the anhepatic phase in the caval replacement group.
Hyperglycemia has been associated with the development of AKI.21 In patients with a Cr concentration less than 1.5 mg/dL, univariate analysis of intraoperative glucose average concentrations was statistically significantly correlated with AKI and continued to demonstrate a strong trend toward a statistically significant correlation (P = 0.07) in the multivariate analysis. This association needs to be validated further in larger studies.
This prospective observational study provides evidence that the risk of developing intraoperative AKI is substantial in patients undergoing liver transplantation.
In patients with a Cr concentration of less than 1.5 mg/dL, the incidence of renal dysfunction as evidenced by elevated Cr levels was greater than 50%. Further analysis demonstrated that 15 patients (33%) fulfilled the RIFLE criteria for renal injury (an increase in Cr by a factor of 2 in comparison with baseline values). Our results are similar to data published by O'Riordan et al.22 These findings gain further importance in the context of recent epidemiological data indicating that a Cr increase as small as 0.3 mg/dL in hospitalized patients carries a significant risk for adverse outcomes and a 6 times increased odds ratio of mortality.6 The importance of these results was confirmed in liver transplantation patients by Barri et al.,11 who showed that hospital-acquired AKI in this patient population was associated with increased morbidity and reduced graft and patient survival.
In the past, renal function in liver transplantation was predominately seen in the context of preoperative risk stratification based on existing renal dysfunction2 and a postoperative decline in renal function resulting in chronic kidney disease or end-stage renal disease (ie, induced by immunosuppression).5
The intraoperative period and its role in AKI were neglected up to recently, in part because of the lack of adequate and timely surrogate markers. To complicate matters, commonly used endpoints may have been confounded by other variables that are not directly related to the intraoperative course (ie, postoperative hypotension in the intensive care unit). Furthermore, the interpretation of commonly used endpoints such as Cr remains difficult, and Cr is known to overestimate renal function in patients with end-stage liver disease. Similarly, the estimated GFR may not be accurate in the setting of dynamic changes in renal function.
This study provides preliminary evidence that intraoperative serum NGAL levels may be a suitable surrogate marker for AKI in liver transplantation. The observed NGAL profile may better identify risk for the perioperative development of AKI in the operating room.
NGAL has been successfully investigated in cardiac surgery, critical care, and kidney transplantation. Although a direct comparison of absolute NGAL concentrations between different studies remains difficult because of the lack of standardized assays for NGAL, we could confirm that a single NGAL determination as early as 2 hours (NGAL2) after an injury (in our case the anhepatic phase) is correlated with postoperative AKI in patients with Cr concentrations less than 1.5 mg/dL. Baseline NGAL, which possibly may detect subclinical renal injury, was not associated with the postoperative development of AKI in patients with preserved renal function, although our cohort may not be sufficiently powered to detect smaller differences. The delta of NGAL2 − NGAL1 was predictive for AKI when all patients (including patients with preexisting renal dysfunction) were included in the analysis. NGAL2 alone was not significant because patients with a high concentration (Cr ≥1.5 mg/dL) already had an elevated NGAL concentration at baseline. Confirmation of AKI in its very early or evolving state would be beneficial because this would possibly allow a timely intervention before the injury process is completed and the return of renal function mainly relies on renal repair mechanisms. Our results suggest that this may become possible with newly available biomarkers.
Several studies have retrospectively investigated risk factors and predictors for AKI. Only 1 study retrospectively developed a risk stratification model that subsequently was applied prospectively.13 This study confirmed some of the previously identified risk factors for AKI, whereas other reported variables were not confirmed. Taken together, these studies demonstrate the difficulty of employing a unified set of variables for prediction models across different institutions, practices, and study designs. Similar limitations also applied to our study, which did not control for institutional bias or variables not measured or collected. Furthermore, it was not our primary study aim (and therefore the study was not adequately powered) to identify or confirm previously identified risk factors. Despite the fairly small sample size, our cohort points to significant differences in intraoperative care between our institutions (Tables 2 and 3).
Two particular variables that were associated with the development of risk for AKI deserve special attention: the type of surgical technique for graft implantation and glucose homeostasis.
The role of hyperglycemia has received significant attention in recent years and remains controversial with respect to patient survival in the intensive care unit.21, 23 Relevant to this study are several studies of critically ill patients as well as cardiac and trauma patients that demonstrated fairly consistently increased renal dysfunction with poorly controlled glucose concentrations.23–30 Our group also demonstrated in a large retrospective study that glucose concentrations are associated with worsening renal function in organ donors.31 This study prospectively demonstrates a strong trend in the multivariate analysis between glucose homeostasis and the development of AKI. However, this finding needs to be validated in studies specifically addressing this question.
The type of surgical technique was also highly correlated with the development of AKI. In our cohort, the mean blood pressure during the anhepatic phase was not different between caval replacement with bicaval anastomoses and piggyback techniques. Complete occlusion of the inferior vena cava results in a renal outflow obstruction. Experimentally, renal vein outflow obstruction has been shown to cause severe renal injury despite preserved arterial renal inflow.32 The present data suggest that maintenance of partial flow, by either a piggyback technique or possibly the use of venovenous bypass, may protect the kidneys during the anhepatic phase, which is not entirely a function of blood pressure. The type of surgical technique should perhaps be considered in select patients with preexisting renal dysfunction. As stated earlier, the interpretation of this finding has to be limited because the type of surgery was not randomized but depended on institutional preference and we did not have sufficient data to separately analyze patients with baseline Cr concentrations greater than 1.5 mg/dL (n = 14). Other variables not accounted for may have contributed to these findings.
One incidental result deserves some explanation. When all patients were considered, baseline Cr was significantly lower in the group with AKI (1.0 ± 0.4 versus 1.5 ± 0.9 mg/dL). This difference disappeared in the subset analysis of patients with baseline Cr < 1.5 mg/dL. This appears to suggest that patients with better renal function are at greater risk of AKI. We believe that this is an artifact of the definition of AKI, which is a 50% or greater increase. An examination of Cr kinetics indicates that for the same degree of renal injury, Cr would not increase to the 50% increase threshold as quickly for a patient with a high baseline Cr concentration.20
In conclusion, this study establishes NGAL as a potentially useful marker for the detection of early AKI and underscores the importance of systematically studying intraoperative AKI during liver transplantation. This special patient population, for a variety of reasons, poorly tolerates an additional insult to the kidney, which may have a detrimental impact on short-term and long-term outcomes. Newly established markers for injury, such as NGAL, in conjunction with improved markers for renal function will allow us to further delineate the natural course of AKI during liver transplantation. This in turn will direct further interventional trials during the perioperative period.