Acute Kidney Injury and Renal Regional Oxygen Saturation During Pediatric Liver Transplantation

Background

Acute or chronic liver failure in children affects various organs, among which, acute kidney injury (AKI) is a potentially serious complication. Pediatric liver transplantation (pLTx) is associated with hemodynamic and procedural alterations and thus can aggravate AKI in children undergoing pLTx. Although there have been few studies focussed on these patients, some studies showed that the incidence of AKI is about 5.2% and increases postoperatively to up to 46% [1].

Near-infrared spectroscopy (NIRS) technology uses an infrared light source to measure regional oxyhemoglobin (SrO2) saturation non-invasively, continuously, and in real-time [2].Several studies have been published on infants and children who underwent cardiac surgery with cardiopulmonary bypass, showing that changes in renal SrO2 (rSrO2) are significantly correlated with development of postoperative AKI [3–7].

We hypothesized that intraoperative NIRS is a helpful tool to detect preoperative renal failure and that the intraoperative decrease in rSrO2 is associated with postoperative kidney injury. Therefore, we retrospectively analyzed intraoperative rSrO2 in children undergoing pLTx and compared the rSrO2 with the incidence of pre- and postoperative AKI.

Material and Methods

PATIENTS:

This study was conducted in agreement with the regulations of the Ethics Committee of the University of Regensburg (approval no. 16-101-0005) and according to the ethics guidelines of the 1975 Declaration of Helsinki. Infants and children who underwent liver transplantation for acute or chronic liver failure at our clinic between March 2011 and March 2016 were included. Patients received no oral premedication and were fasted according to the local protocol. The initial monitoring included electrocardiogram, non-invasive blood pressure, and pulsoxymetric saturation (SpO₂). Anesthesia was induced with sufentanil (0.2 μg/kg per BW), propofol (2–4 mg/kg per BW), and rocuronium (0.5 mg/kg per BW). After intubation, desflurane 0.6–0.9 MAC (in 30–50% O₂/air mixed to keep the paO₂ >100 mmHg) was used for maintaining anesthesia. Gas monitoring for desflurane was performed and end-tidal CO₂, inspiratory and expiratory oxygen concentration, as well as spirometry (gas flows, volumes, and pressures), were measured. Body temperature was measured in the oesophagus and in the bladder via a urinary catheter. In all children, invasive arterial blood pressure, continuous central venous pressure, and urinary output were monitored. Blood for arterial blood gases were taken after the induction of anesthesia, prior to reperfusion, approximately 5 min after reperfusion of the new graft and whenever indicated by the anesthesiologist. The concentrations of hemoglobin, electrolyte, lactate, and glucose and arterial blood gases were measured and recorded. Blood samples to determine laboratory values (for creatinine, cystatin C, and bilirubin) were taken prior to transplantation, on arrival in the ICU, and on the following 7 days in the morning. For volume replacement, the children received crystalloid (5–15 mg/kg/h) and colloidal solutions. Blood transfusion for maintaining the hemoglobin concentration above 7 mg/dl and fresh frozen plasma were transfused when thromboplastin time was below 50% and pooled thrombocyte concentrate in the case of a thrombocyte count below 60 000/μl.

CREATININE AND CYSTATIN C CONCENTRATION:

Creatinine values were taken from the laboratory results of the patients. Creatinine concentration threshold was determined according to the method described by Lechner and Liebau [8]. According to the local protocol, serum creatinine concentrations were determined preoperatively and once every postoperative day. For analysis, preoperative and highest concentrations during the first postoperative creatinine values of the first 7 postoperative days were used.

Normal values for cystatin C concentration were determined according to the method described by Ziegelasch et al., and the 50th percentile was used as the threshold value [9]. According to the institutional standards, serum cystatin C concentrations were determined preoperatively and on postoperative day 7.

Urea nitrogen concentrations were determined preoperatively and every day following pLTx; according to the references our laboratory, threshold values were set at 48 mg/dl.

PEDIATRIC RIFLE CRITERIA:

KI was defined using the pediatric RIFLE (pRIFLE) criteria for acute kidney injury. This creatinine-based definition is determined by glomerular filtration rates or reduction in urine output. The GFR was calculated using the Schwartz formula. AKI is categorized into stage I (risk), stage II (injury), and stage III (failure). These stages correspond with a decrease in the GFR of 25%, 50% or 75%, respectively, or a decrease in urine output [10].

CHRONIC RENAL FAILURE:

Chronic renal failure was defined when children had increased cystatin C or creatinine concentrations in the 3 months prior to LTx.

NEAR-INFRARED SPECTROSCOPY MEASUREMENT:

Following induction of anesthesia, the left kidney was identified using ultrasound, and a pediatric (<40 kg BW) or adult (>40 kg) SrO₂ electrode (pediatric or adult INVOS™ system oximetry sensor, Covidien, Minneapolis, MN, USA) was placed on the back over the left kidney. rSrO₂ was continuously measured using the INVOS 5100B oximeter sensor system (Covidien, Minneapolis, MN, USA). Measurements were recorded throughout the pLTX until the patient was transferred to the ICU. The 5100B cerebral oximeter generates low-intensity near-infrared light and directs the light onto the patient’s skin at wavelengths of 730 and 810 nm. The light penetrates the skin and passes through the tissue. The sensor of the oximeter consists of a light-emitting diode and 2 detectors located at a distance of 3 and 4 cm from the diode.

DATA COLLECTION:

All demographic patient data and laboratory values were taken from the patients’ records. Information on drugs and laboratory values were recorded prior to and during the first 7 postoperative days. The hemodynamic parameters were taken from the anesthesia charts. Data were recorded 45 min after the induction of anesthesia, 45 min after the start of the anhepatic phase, and 45 min after reperfusion. Laboratory values were taken from the patients’ electronic records.

rSrO₂ data were collected automatically every 5 to 6 s and stored in 2.5-min intervals on a computer hard drive for later analysis. For statistical analysis, the mean of rSrO₂ was used. A decrease in rSrO₂ was defined as a reduction of the rSrO₂ for more than 20% for more than 10 min compared to the preparation period. Hemodynamic data and blood gas analysis were taken from the anesthesia protocol. Hemodynamic data and SrO₂ were analyzed every 5 min.

STATISTICAL ANALYSIS:

Statistical analysis was performed with SPSS software version 24 (SPSS, Inc., Chicago, IL, USA). The relationship between SrO₂, rSrO₂, MAP, creatinine, cystatin C, bilirubin, and other laboratory variables was assessed using Pearson’s correlation coefficient. Data were tested for normal distribution using the Kolmogorov-Smirnoff test. Comparisons of the means of the 2 groups were conducted after the Levène test by using paired or unpaired t tests where necessary. Kruskal-Wallis test with Bonferroni correction or χ² test were performed to assess significant differences between groups. Stepwise logistic regression was conducted to evaluate the influence of bilirubin on NIRS measurement and renal failure. A p value <0.05 was regarded as statistically significant.

Results

:

The rSO2 measurements were affected by several factors (Table 3).

PELD SCORE:

The PELD score inversely correlated with rSrO₂ (r=−0.378, p=0.05), indicating that sicker children presented with lower rSrO₂.

LABORATORY VALUES: Several laboratory values affect the NIRS technology. Correlations are summarized in Table 3. Hemoglobin levels significantly affect NIRS measurements (r=0.371, p<0.001). Lower hemoglobin levels were associated with decreased rSrO2 measurements. Blood gas parameters, which are associated with renal perfusion and oxygen supply, revealed that rSrO2 was significantly dependent on paO2.

When comparing children with increased pre- or postoperative cystatin C levels, there was no significant increase in duration of transplantation, anhepatic period, days on ventilation, or ICU stay between groups (Table 4). However, when defining kidney injury by increased serum creatinine concentration, we detected a significant difference in the duration of ICU stay in children with and without kidney injury (Table 5).

BILIRUBIN:

Bilirubin interferes with NIRS measurement. In our study population, the total bilirubin levels dropped significantly from 11.4±9.2 to 4.5±4.1 (r=0.994, p<0.001) during liver transplantation. The level of bilirubin significantly affected the rSrO2 measurement (Table 3).

:

Figure 1 indicates the cumulative results of the temporal course in all 41 children undergoing pLTx (Figure 1). The rSrO2 did not significantly differ between the preparation period (61.1±21.8) and the anhepatic phase (62.4±18.9, p=0.253) but significantly increased in the reperfusion period (65.2±20.7, anhepatic period vs. reperfusion: p<0.001) compared to the preparation period.

MEAN ARTERIAL PRESSURE:

Mean arterial pressure was significantly correlated with the renal NIRS measurement (r>0.5, p<0.01) except for the first 10 min following reperfusion (r=027, p=0.165) (Figure 1). MAP increased significantly but without correlation with the rSrO2 measurement.

:

Fourteen out of 41 children presented with increased preoperative cystatin C levels. Children with increased cystatin C concentrations had significantly lower rSrO2 during pLTx than children without (63.7±4.3 vs. 53.4±4.9, p<0.05) (Figure 2A). Stepwise logistic regression revealed that the significant difference in rSrO2 was not due to increased serum bilirubin, hemoglobin concentration, or FiO2 in these children (cystatin C: regression coefficient −9.979 SE: 5.9, OR: −21.8–1.9 p=0.096, total bilirubin: regression coefficient: −0.557, SE: 0.322, OR: −1.21–0.096, p=0.092. Hb: regression coefficient: 6.599 SE: 0.94, OR: 4.673–8.525, p=0.01, FiO2: regression coefficient: 0.239, SE: 0.07 OR: 0.095–0.382 p=0.106). Before pLTx, serum creatinine concentration was pathologically increased in 4 out of 41 children. There was no statistically significant difference in rSrO2 in children with normal vs. increased serum creatinine concentrations (Figure 2B).

:

Serum cystatin C concentration did not change significantly between preoperative and postoperative values on day 7 (preoperative cystatin C serum concentration was 1.01±0.4 mg/dl and postoperative concentration was 1.03±0.3 mg/dl). We could not detect any significant difference between children with (n=22) preoperative or postoperative elevated (n=18) cystatin C concentrations and the rSrO2 (Figure 3A).

In the study group, 17 out of the 41 children developed kidney failure according to the increase in serum creatinine concentration during the first week after pLTx. There was a tendency to increased rSrO2 in children with postoperative AKI, but this did not reach statistical significance (Figure 3B).

:

Children developed renal failure during the first 7 days following transplantation, with a peak on postoperative day 7. During the postoperative period, 11 children developed stage I renal failure according to the pRIFLE criteria, 1 child developed stage II, and 6 children stage III, and 23 children no renal failure. There was a significant difference between the 4 RIFLE stages on postoperative days 2–5 (p<0.05). However, there was no significant difference in intraoperative rSrO2 in children who developed renal failure and those without. A significant correlation between the length of ICU stay (p=0.038) and the development of pRIFLE stage I to III was detected. However, no significant correlation could be shown between PELD score, the duration of the transplantation or mortality, and the development of kidney failure (Figure 4).

:

Significant alterations in rSrO₂ might be associated with postoperative compromised renal function. We examined whether the increase or the drop in rSrO₂ for 25% or more for more than 10 min was associated with the worsening of kidney function, defined by an increase in serum cystatin C, creatinine concentration, or pRIFLE criteria. There was no significant increase or decrease in worsening of renal function in children who experienced a drop or increase in rSrO₂ of 25% or more (cystatin C, n=10, p=0.421, creatinine concentration, n=6, p=0.477).

Discussion

Renal failure is common in children undergoing pLTx. Due to delayed response in kidney injury, currently available biomarkers provide only delayed information about kidney function. Here, we show that rSrO2 was correlated with renal failure defined by increased preoperative serum cystatin C concentration but not serum creatinine concentration. In addition, the intraoperative increase or decrease in rSrO2 was not associated with the development of postoperative kidney injury defined either by the postoperative increase in serum creatinine concentration, cystatin C concentration, or the pRIFLE criteria. Renal NIRS might provide additional information on renal function during pLTx.

Renal failure is a common complication in patients with acute or chronic liver disease. In contrast to conditions with multiorgan involvement, like the Alagille Syndrome or metabolic disorders, most children with end-stage liver disease develop AKI secondary to the chronic liver disease they originated during infancy [11,12].

The incidence in children with liver disease is reported to be about 20% [13].

There are several definitions of renal failure in children, including pediatric RIFLE criteria (Pediatric Risk, Injury, Failure, Loss of function and End-stage renal disease), AKIN (Acute Kidney Injury Network), and KDIGO (Kidney Disease for Improving Global Outcomes) [10]. All clinically relevant definitions of AKI are based on serum creatinine and urine output. However, both parameters are underestimated in children with liver failure [14]. In addition, the definitions for AKI were not developed for patients with liver failure and lack standardization and sensitivity [15]. More than 50% of the glomerular function is thought to be lost before the diagnosis of AKI is considered, and this affects prognosis because possible therapy is usually delayed [16]. Cystatin C is a newer biomarker for kidney injury. The concentration is independent of muscle mass and it was shown that it is a reliable marker for assessment of renal dysfunction in children with liver disease and after LTx [17]. In a meta-analysis, cystatin C was reported to have acceptable prognostic value for the prediction of AKI in children [18]. Due to metabolism and reabsorption, urea nitrogen is a weak parameter for renal failure.

NIRS technology enables non-invasive and real-time measurement of tissue oxygenation. In pediatric cardiac anesthesia, NIRS allows continuous monitoring of non-invasive organ-specific perfusion. Particularly, regional cerebral oxygen saturation (cSrO2) in infants and neonates is now widely accepted as a standard monitoring tool to detect oxygenation. Perfusion mismatch and low cSrO2 values were associated with poorer patient outcome or longer hospitalization [19]. NIRS technology is affected by various laboratory parameters. As shown previously for cSrO2, total and unconjugated bilirubin significantly influence rSrO2 during LTx [20,21]. rSrO2 is significantly correlated with the hemoglobin concentration. In addition, we detected a significant positive correlation between serum Na+ and Cl− concentrations and a negative correlation for K+ and cSrO2. These electrolytes may represent indirect parameters for the severity of liver disease resulting in skin edema formation or renal dysfunction [22,23]. We detected a significant correlation between rSrO2 and the PELD scores in our study population. The PELD score reflects waiting time and severity of liver disease in children up to 12 years of age. The calculating factors include age, international normalized ratio, bilirubin and albumin concentrations, a history of growth failure, and time on the waiting list.

In our study, we could not find a correlation between markers of renal failure and rSrO2. In contrast to other studies, we did not detect an association between the intraoperative increase or decrease of 25% or more in rSrO2 compared to baseline and postoperative renal failure according to the pRIFLE criteria, the serum creatinine, or the cystatin C concentration [24].

Several studies in pediatric cardiac surgery in which the correlation of rSrO2 with postoperative AKI revealed only inconsistent results, which might be explained by the different definitions for AKI used in the studies [3–6]. When using the pRFILE criteria to define AKI, 2 studies demonstrated a significant correlation between intra- or postoperative AKI and a decrease in rSrO2 compared to children with normal kidney function [3,4,6]. Hazle et al. found a significant correlation of AKI defined by various urinary biomarkers and the cumulative time of NIRS < 50% [4]. However, in children following aortic arch repair, no significant correlation between the development of AKI and rSrO2 was found [5]; For the estimation of creatinine concentration, they also used the Schwartz formula and determined the threshold according to the pRIFLE criteria. In a recently published trial, serum creatinine concentrations were used to define AKI according to the pRIFLE criteria in children undergoing cardiac surgery with cardiopulmonary bypass. As in our study, when comparing rSrO2 with the development of renal failure, they also detected an increased rSrO2 in the children who presented with postoperative renal failure [7]. In children undergoing kidney transplantation, postoperative renal rSrO2 assessed by NIRS strongly correlates with common markers of kidney graft function and perfusion, allowing continuous real-time monitoring of blood flow in renal grafts by NIRS [25].

There are several definitions of AKI in children [10], and the incidence of AKI depends on the definition of AKI [10]. Serum cystatin C concentration seems to be a more reliable indicator, especially in children with liver failure, than serum creatinine-derived values [16,17,26]. Additional, there are several other reasons for the development of renal failure in children undergoing pLTx, including immunosuppression (e.g., with ciclosporin which has profound effects on renal function) [27].

The present study has several limitations. Due to its retrospective design, postoperative cystatin C values were only available on the 7th day after liver transplantation. We did not determine edema and could not determine the impact of skin perfusion or the impact of invasive ventilation on NIRS measurement.

Conclusions

In this retrospective study, we show that decreased rSrO2 during pLTx indicates compromised renal function as represented by increased serum cystatin C concentration. rSrO2 did not correlate with preoperative serum creatinine concentration. In addition, intraoperative renal NIRS did not predict postoperative renal failure, indicated by increased postoperative serum cystatin C, serum creatinine concentration, or pRILFE. Renal NIRS indicated compromised renal function and can be included in pLTx. The role in therapeutic interventions needs to be evaluated in prospective studies.