Factors affecting serum concentration of vancomycin in critically ill oliguric pediatric patients receiving continuous venovenous hemodiafiltration

Vancomycin is known to be unintentionally eliminated by continuous renal replacement therapy, and the protein bound fraction of vancomycin is also known to be different in adults and children. However, there are only a few studies investigating the relationship between the dose of continuous venovenous hemodiafiltration (CVVHDF) parameters and serum concentration of vancomycin in pediatric patients. The aim of this study was to determine clinical and demographic parameters that significantly affect serum vancomycin concentrations. This retrospective cohort study was conducted at a pediatric intensive care unit in a tertiary university children’s hospital. Data from oliguric patients who underwent CVVHDF and vancomycin therapeutic drug monitoring were collected. The correlation between factors affecting serum concentration of vancomycin was analyzed using mixed effect model. A total of 177 serum samples undergoing vancomycin therapeutic drug monitoring were analyzed. The median age of study participants was 2.23 (interquartile range, 0.3–11.84) years, and 126 (71.19%) were male patients. Serum concentration of vancomycin decreased significantly as the effluent flow rate (EFR; P < 0.001), dialysate flow rate (DFR; P = 0.009), replacement fluid flow rate (RFFR; P = 0.008), the proportion of RFFR in the sum of DFR and RFFR (P = 0.025), and residual urine output increased. The adjusted R2 of the multivariate regression model was 0.874 (P < 0.001) and the equation was as follows: Vancomycin trough level (mg/L) = (0.283 × daily dose of vancomycin [mg/kg/d]) + (365.139 / EFR [mL/h/kg])–(15.842 × residual urine output [mL/h/kg]). This study demonstrated that the serum concentration of vancomycin was associated with EFR, DFR, RFFR, the proportion of RFFR, and residual urine output in oliguric pediatric patients receiving CVVHDF.


Introduction
Vancomycin is the first-choice drug for methicillin resistant Staphylococcus aureus infections, and is one of the most frequently used antibiotics in the intensive care unit (ICU) [1,2]. Since a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 vancomycin is excreted not only through the kidneys but also through continuous renal replacement therapy (CRRT), studies have reported that the serum concentration of vancomycin is affected in patients receiving CRRT [3][4][5][6][7]. In addition, serum vancomycin concentrations are known to be affected by the dose of CRRT [8].
Maintaining an adequate level of antibiotic in patients with infectious diseases is undoubtedly important. Failure to maintain adequate serum levels results in treatment failure or drug toxicity, and for critically ill patients, the outcome may be fatal [6,7,9]. Considering that most patients that are administered with vancomycin whilst receiving CRRT are critically ill, maintaining therapeutic serum concentrations of vancomycin is one of the most important factors related to successful treatment outcome in these patients. For this reason, several studies have been conducted on the pharmacokinetics and the optimal dose of vancomycin during the application of CRRT [6,[10][11][12][13][14][15]. However, most studies have focused on the effects of either hemofiltration or hemodialysis [10,13]; and there are relatively few studies focused on continuous venovenous hemodiafiltration (CVVHDF) involving the interaction of continuous venovenous hemofiltration (CVVH) and continuous venovenous hemodialysis (CVVHD) [6,11].
A previous study reported that vancomycin clearance was considerably enhanced with hemofiltration compared to hemodialysis [16]. In contrast, another study reported that the difference in vancomycin clearance was insignificant between hemofiltration and hemodialysis [17]. Another study recommended higher vancomycin doses in patients receiving CVVHD than CVVH, suggesting that vancomycin clearance by hemofiltration was lower than by hemodialysis [12,15]. Regardless, all of these studies were conducted in adults. Since the protein bound fraction of vancomycin has been reported to be lower in children than in adults [18,19], there is a limit to applying the results from adult studies directly to children.
Therefore, this study aimed to analyze the effect of the dose of each CVVHDF parameter and clinical characteristics on the serum concentration of vancomycin in pediatric patients receiving CVVHDF.

Study population
This retrospective cohort study was conducted at a pediatric ICU (PICU) with 24 beds, in a tertiary university children's hospital. Patients 18 years old who received CVVHDF and were subjected to vancomycin therapeutic drug monitoring (TDM) at the PICU between January 1, 2005 and August 31, 2015 were included. Samples of vancomycin TDM which the steady-state level of vancomycin was not reached during CVVHDF application were excluded from the study; the steady state was expected to be reached after four or more doses of vancomycin were administered at equal intervals, in accordance with the Infectious Diseases Society of America guideline [20]. Only samples collected for TDM 30 minutes before the administration of vancomycin were included to obtain consistent results [20,21]. Because extracorporeal membrane oxygenation (ECMO) has significant effects on the hemodynamics of patients and can change the serum concentration of vancomycin, patients receiving ECMO were excluded. Patients without oliguria (urine output ! 0.5 mL/kg/h) were also excluded because residual renal function in these patients may enhance the clearance of vancomycin [22].

Data collection and definition
Data on patients' age, sex, serum concentration of vancomycin, daily dose and dosing interval of vancomycin, blood flow rate (BFR), dialysate flow rate (DFR), patient fluid removal rate (PFRR), replacement fluid flow rate (RFFR), vital signs, fluid input/output balance, residual urine output, and laboratory test results were collected from the electronic medical records.
All CVVHDF parameters within 24 hours prior to TDM, and if there were more than 2 values for each of the parameters, their mean value was adopted. Effluent flow rate (EFR) was defined as the sum of DFR, RFFR and PFRR; and the proportion of RFFR was defined as the proportion of RFFR in the sum of DFR and RFFR.

Statistical analysis
The correlation between each factor and the serum concentration of vancomycin was analyzed using univariate mixed effect model regression analysis, with patients as random variables and serum concentration of vancomycin as fixed variable. Multivariate regression analysis was performed on the factors that showed significant results in the univariate regression analyses, and the model was determined by backward selection method. In the multivariate regression model, adjustment of the statistical significance was made using Bonferroni correction, and the overall type I error rate was 0.05. All statistical analyses were performed using R version 3.4.3 (R Foundation for statistical computing, Vienna, Austria).

Ethics statement
The Seoul National University Hospital Institutional Review Board approved the study protocol (H-1508-138-697) and waived the need for written informed consent.

Baseline characteristics of patients
Patients who underwent vancomycin TDM while receiving CVVHDF at the PICU during the study period were screened. A total of 177 samples of vancomycin TDM from 75 patients were included in the final analyses based on the inclusion and exclusion criteria (Fig 1). The median age was 2.23 (interquartile range, 0.3-11.84) years old, and 126 (71.19%) samples were obtained from male patients. Table 1 shows the baseline characteristics of the samples included.
Multivariate analysis was performed on the significant variables found in the univariate analysis. The multivariate model derived from backward selection had four predictor variables including daily dose of vancomycin (P < 0.001), dosing interval of vancomycin (P < 0.001), EFR (P = 0.019), and residual urine output (P < 0.001). However, the P-value of dosing interval of vancomycin was greater than the corrected P-value 0.0125 through the Bonferroni method, and the final model was as follows (adjusted R 2 = 0.874, P < 0.001) (

Vancomycin trough level group comparison
The samples of vancomycin TDM were divided into 4 groups (less than 10 mg/L, between 10-15 mg/L, between 15-20 mg/L, and 20 mg/L or more) according to their vancomycin trough levels. The doses of EFR (P = 0.001), DFR (P = 0.031), RFFR (P = 0.017), and residual urine output (P = 0.014) were significantly lower in the groups with a higher vancomycin trough level. On the contrary, PFRR showed no significant difference according to the groups ( Table 3, Fig 3A and 3B).

Discussion
This study was designed to analyze clinical and demographic parameters, to observe which parameters affect the serum concentration of vancomycin. The data analyzed from 177 pediatric samples of vancomycin TDM over a ten-year period demonstrated that DFR, RFFR, EFR, the proportion of RFFR, and residual urine output were significantly related to the serum concentrations of vancomycin.
In this study, the serum concentration of vancomycin decreased with increasing EFR, DFR, and RFFR. This is consistent with previous researches showing that vancomycin is excreted via hemofiltration and hemodialysis, and that the intensity of RFFR and DFR correlated with the clearance of vancomycin [5,7,13,[23][24][25][26]. It is also consistent with previous studies that show EFR to be a reliable predictor of antibiotic concentration in patients receiving CVVHDF [8,27]. In the past, due to the use of low-flux hemofilter, drug clearance by CRRT, especially hemodialysis, was influenced by the size of drug molecule. However, in the recent years, highflux hemofilters have replaced low-flux hemofilters, therefore having almost no influence on drug clearance according to the size of the drug molecule [3,13]. Because all samples of vancomycin TDM included in this study received CVVHDF with a high-flux hemofilter AN69 [28,29], it is possible that vancomycin could be effectively excreted by hemodialysis.
As the proportion RFFR increased, that is, the ratio of hemofiltration to hemodialysis increased, the serum concentration of vancomycin decreased significantly. This is in agreement with studies reporting that vancomycin is more effectively excreted by hemofiltration than by hemodialysis [16], but contrasts studies that recommend a higher vancomycin dose for patients receiving CVVHD or CVVHDF than CVVH [12,15]. Because many factors related to the pharmacokinetics of vancomycin interact with each other in critically ill patients receiving CRRT [30], convection and diffusion may act in synergy with each other, causing an improved convection and over diffusion [11]. However, it is difficult to make a simple comparison with these studies, and in addition, the aforementioned studies were conducted on adults. It is known that the protein binding faction of vancomycin is significantly lower in children than in adults [18,19]. Therefore, the possibility that vancomycin clearance by CVVHDF may be affected by the difference in the pharmacokinetic characteristics of vancomycin in these children compared with adults should be considered carefully.
In this study, to diminish the effect of residual renal function on vancomycin trough level, samples from patients with a urine output ! 0.5 mL/kg/h were excluded from the analyses. Nevertheless, even with negligible residual urine output of <0.5 mL/kg/hr, this study showed that vancomycin trough levels can be significantly affected, which is consistent with previous studies [31,32]. More than a quarter of the samples were in the subtherapeutic range (<10mg/ L [33,34]) even if the administered median vancomycin dose in this patient group was 20.11 mg/kg/day; which is the recommended dose (20 mg/kg/day) for patients receiving CVVHDF. This phenomenon can be explained by vancomycin being excreted not only by CVVHDF, but also by residual diuresis. In fact, patients with a higher residual urine output had lower vancomycin trough levels, moreover, most of these patients were in the subtherapeutic group.
Although vancomycin clearance in patients receiving CVVHDF appears to be easily predicted by the known principles of hemofiltration, hemodialysis, and pharmacokinetic properties of vancomycin, it is recommended that TDM be performed instead of relying only on the predicted value because the actual clinically measured value is often different from the predicted value [3,7,8,14,35,36]. In particular, vancomycin has a narrow therapeutic range, thus accurate prediction and maintenance of proper serum concentration are more important [20,21]. For these reasons, the necessity of research based on actual clinically obtained results, as well as experimental research has emerged; and in fact, such clinical studies have been carried out [8,30]. Nevertheless, only a few clinical researches include data on children. This study is significant in that there is no data comparing the proportion of hemofiltration and hemodialysis which affects the serum concentration of vancomycin in children receiving CVVHDF.
This study has several limitations. First, because this study was a retrospective study, controlling all the variables to exclude bias was difficult, and incomplete data restricted the extent of the study. An example being the duration of hemofilter use, which may have affected  vancomycin clearance. However, we were unable to obtain this information. Second, this study derived a model through multiple regression analysis. If population pharmacokinetics were used, a more optimized model could be derived. Finally, this study was conducted at a PICU in a single center, and therefore caution is needed when applying the results of this study to other centers. Nevertheless, data obtained from this study can be the basis for future prospective studies. Also, it is important for bringing awareness to the necessity of studies on children receiving CVVHDF.

Conclusions
In conclusion, this study demonstrated that serum concentration of vancomycin was associated with EFR, DFR, RFFR, the proportion of RFFR, and residual urine output in oliguric pediatric patients receiving CVVHDF. A well-designed multicenter study is needed to draw conclusions that are more specific and applicable at a broader level.
Supporting information S1 Table. Raw data used in this study for statistical analyses can be found in S1 Table. (XLSX)