The Pharmacokinetics , Optimal Dose and Therapeutic Monitoring of Vancomycin in Severe Burn Patients

Objectives: This paper investigated the pharmacokinetic (PK) change, optimal dose and therapeutic monitoring method of vancomycin in severe burns. Method: This is a retrospective, matched cohort study. Nine burn patients were included, and each individual was matched with a control critically ill patient. Vancomycin concentrations were analyzed using the first order pharmacokinetic principles to calculate elimination rate constant (K), half-life (t1/2), volume distribution (Vd), total body clearance (CLvancomycin) and the 24-hour area under the curve (AUC24 h). Population PK analysis and Monte Carlo simulation was used to investigate optimal dose. Results: The mean burn area of the burn group and age was 60% and 20 years-old respectively. The loading dose and daily maintenance dose was significantly higher in burn group than that in control group. Vancomycin clearance was significantly higher (p<0.05) in burns patients when compared to control patients. CLCr (creatinine clearance) was not significantly correlated to CLvancomycin (vancomycin clearance) in both groups. Ctrough (serum trough concentration) was significantly correlated with AUC24h in the control group (r=0.98, R 2=0.96, p<0.01), but not in the burn group (r=0.63, R2=0.40, P=0.07). Daily dose of 5000 mg of vancomycin could achieve 90% probability of target attainment if target was AUC24h/MIC (minimum inhibitory concentration)>400. Conclusion: Two steady state concentrations but not Ctrough is an appropriate reference for vancomycin therapeutic monitoring. A daily dose of at least 5000 mg is suggested in severe burn patients with normal renal function (CLCr>90 mL/min).


Introduction
Vancomycin, a potential nephrotoxic antibiotic, is commonly used in the treatment of methicillin-resistant Staphylococcus aureus (MRSA) infection or sepsis.The antibiotic is excreted by kidney and 55% of it is bound to albumin [1].However, the pharmacokinetics (PK) is significantly altered in severe burn patients [2,3].Previous studies have shown that the renal clearance and volume distribution (V d ) of vancomycin in burns patients is higher than it is in controls, while the half-life is shorter [4][5][6].Thus, larger loading and the use of more frequent dosing of vancomycin especially when intermittent dosing is used are needed to maintain therapeutic serum level, and inadequate dosing is not an uncommon clinical problem [7].
To monitor clinical efficacy, the 24 h area under the serum concentration-time curve to minimum inhibitory concentration ratio (AUC 24h : MIC) of >400 has been suggested as a standard clinical reference for vancomycin dosing [8].As C trough significantly correlates with the AUC 24 h , a C trough level of 15-20 mg/L is recommended to achieve an AUC 24h / MIC of 400 if the MIC is 1 mg/L.However, its use as reference for dosage adjustment in severe burns patients has been questioned as the correlation of C trough with AUC 24h remains debatable in this special group [9,10].
The main purpose of this retrospective study was to investigate the correlation of C trough with AUC 24h in severe burn patients.The burn patients in similar ages and injury mechanism from the 2015 Taiwan cornstarch dust explosion have provided a clinical model with least spatial and temporal effect.Another aim of this study was to define the optimal dose for this population to achieve AUC 24h / MIC>400.The study obtained approval from The NTUH Ethics Committee (NTUH-201508031DINA).

Methods
This is a retrospective chart review study.Thirty-three adult burn patients (>18 years old) admitted to the National Taiwan University Burn centre following the color cornstarch powder explosion in a water park on June 28, 2015 were reviewed.Only those who received vancomycin therapy at least for 2 days after burn injury with peak (C peak ) and trough (C trough ) serum concentrations available were included in the study.
Patient demographic data including age, sex, height, weight, total Page 2 of 7 body surface area of burn (%), timing of vancomycin infusion (day after burn injury), S Cr (serum creatinine), albumin, indications, loading dose, maintenance dose, dosing interval, infusion time and serum concentration were collected.Weight and S Cr were recorded on the day of blood sampling for serum concentration.
The dose and dosing interval were decided by attending physicians or clinical pharmacists, and the infusion time of vancomycin was 1 hour for dose lower than or equals to 500 mg, and 2 hours for dose between 500 to 1000 mg and so on.The C peak C trough were followed at 2 hours after the end of vancomycin infusion, and within 1 hour before vancomycin infusion.Both were followed after at least 5 doses of vancomycin.Vancomycin concentrations were analyzed by using first order pharmacokinetic principles (Sawchuk and Zaske method) to calculate PK parameters such as elimination rate constant (K), halflife (t 1/2 ), V d , total body clearance (CL vancomycin ) and AUC 24h [11].The formula was shown as follows: For each patient in the burn group, a control subject in intensive care unit was matched by creatinine clearance (CL Cr , mL/min).Matching was based on CL Cr value with a difference of<30 mL/ min between matched pair, or on different stages of chronic kidney disease with a range of CL Cr value between groups (> 90, 60-89, 30-59, <30 mL/min) [12].The CL Cr was estimated by the Cockcroft and Gault (CG) method [13].If patient's S Cr was lower than 1.0 mg/ dL, it will be replaced by 1.0 mg/dL for better estimate of patient's renal function in intensive care unit [14].For obese patients ( ) -/ 30%) true body weight ideal body weight ideal body weight >     the ideal body weight was used to estimate the CL Cr .
The vancomycin serum concentrations were measured by a fluorescence polarisation immunoassay (AxSYM ® system, Abbott Laboratories, Abbott Park, Illinois, USA).The sensitivity of the assay (the lowest measurable level) was 2 mg/L, and the coefficients of variation at 7.0, 35.0 and 75.0 mg/L were all less than 5%.
For optimal vancomycin dose investigation, the population pharmacokinetics analysis was performed for burn patients by NONMEM version 7.2 (ICON Development Solutions, Ellicott City, MD, USA) with G77 FORTRAN compiler.Both one and twocompartment structural models were tested using the first-order conditional method of estimation (FOCE).Clearance and volume of distribution were estimated.A proportional error model was used to characterize the inter-individual variability (% CV) of each parameter.Different residual error models (i.e., additive, proportional and combined additive and proportional) were examined in order to find the better error model to characterize the unexplained residual variability.Due to the limited sample size (n=9), no attempts were made to identify the potential influential covariates.).The final model estimates were compared to the median and the 2.5 th and the 97.5 th percentiles of the parameter estimate obtained from successfully converged bootstrapping runs.The final population PK model was used for simulations to investigate the probability of target attainment (PTA) for different daily dose (2000 mg to 6000 mg).The pharmacodynamic target attainment was defined as achieving AUC 24h /MIC ≥ 400, and the MIC range values were 0.125, 0.25, 0.5, 1.0 and 2.0 mg/L [8].For each dose examined, 5,000 burn patients were simulated using Monte Carlo simulation in NONMEM and the percentage of patients exceeding an AUC 24h /MIC of 400 was computed.
Continuous data were summarized as mean with standard deviation.The demographic and PK data were compared by a twotailed paired t test.Simple linear regression analysis was used to test the correlation between variables.A p-value of 0.05 or lower was considered statistically significant.The statistical analysis was performed using SPSS 11.5 (SPSS Inc, Chicago, Illinois, USA).

Results
Nine young adult burn patients were included in this study, and each individual was matched with a non-burn critical ill patient (control group).The mean surface area of second-and third-degree burn was 60% TBSA (total body surface area).There were no significant differences between the 2 groups in the demographic data except age and albumin (Table 1).Patients with burn injury were younger than the non-burn group (20.78 vs. 33.89years-old) and had lower albumin level (2.6 vs. 3.1 g/dL).All patients in the burn group received vancomycin for the treatment of MRSA empirically.Vancomycin was given before the 10 th day after burn except 1 started on 27 th day.The indications of vancomycin for the non-burn group were MRSA related bacteremia, pneumonia and wound infection in 3 patients respectively, and empirical use for the others.
In the burn group, the loading dose and daily maintenance dose was 25.42 ± 4.53 mg/kg and 45.03 ± 10.43 mg/kg/day respectively, which was significantly higher than that of non-burn group (14.46 ± 1.40 mg/kg and 30.53 ± 12.86 mg/kg/day) (Table 2).The vancomycin clearance was significantly higher in the burn group than that in the non-burn (10.66 L/hr ± 3.28 vs. 6.85L/hr ± 3.54), corresponding to a 55% higher CL in burn patients (p<0.05).Similarly, the trend of K and V d , was higher in the burn group than the non-burn although it was statistically insignificant.CL Cr was not significantly correlated with vancomycin CL in both groups (r=0.13,R 2= 0.02, P=0.74 in burn group vs. r=0.59,R 2= 0.35, P=0.09 in non-burn group) (Figure 1).C trough was not significantly correlated with AUC 24h in the burn group (r=0.63,R 2= 0.40, P=0.07), but was correlated significantly in the non-burn group (r=0.98,R 2= 0.96, p<0.01) (Figure 2).
A one-compartment model with first-order elimination adequately characterized the observed concentration time profiles of vancomycin in burn patients.The final population pharmacokinetic model resulted in a clearance of 9.56 L/hr with an inter-individual variability of 17.3%.Volume of distribution is determined to be 50 L with an inter-individual variability of 37.4% (Table 3).Residual variability was best described by an additive model and was determined to be 0.115 mg/L.The diagnostic plots of the final model are given in Figure 3. Overall, there was a good correlation between observed and individualand population-predicted vancomycin concentrations (Figures 3A and  3B).No obvious bias pattern was observed for the plots of CWRES versus time (Figure 3C) and IWRES versus individual predicted concentrations (Figure 3D).From 200 bootstrap runs, 180 runs with successful minimization were included in the bootstrap analysis.All model estimates were similar between the bootstrap analysis and the final model.Overall the bootstrap analysis indicated that the final model was stable and estimated parameters were reasonably precise.

Page 4 of 7
The PTA (AUC 24h /MIC > 400) for different daily dose of vancomycin by MIC was presented in Figure 4. Based on the simulation results, it was observed that daily dose greater than 3000 and 5000 mg yielded PTA > 90% against MIC of ≤ 0.5 and ≤ 1.0 mg/L, respectively.However, at a MIC of 2.0 mg/L, even though increasing the daily dose up to 6000 mg only provided a PTA of 10%.

Discussion
MRSA infection or sepsis is a life-threatening clinical problem commonly encountered in severe burns.A potential nephrotoxic antibiotic, vancomycin, is usually the treatment of choice.The outcome is associated with proper antibiotic dosing.Inadequate doses will lead to suboptimal bactericidal activities leading to mortality or the development of resistance, while overdoses will cause nephrotoxicity.C trough of 15-20 mg/L and AUC 24h /MIC>400 have been recommended.
As in other studies in the literature, this study has shown that the clearance of vancomycin is augmented in severe burn injury [4][5][6].The results provide an explanation for the observation that although burn patients received higher loading and maintenance dose of vancomycin, the C trough was not significantly higher than that in the non-burn group.Increased glomerular filtration rate, extensive loss through burn wounds, and decreased or altered serum binding proteins are possible explanations for the enhanced vancomycin elimination noted in the burn patients [2].A study of serum protein-binding characteristic of vancomycin has shown that vancomycin preferentially binds to albumin and immunoglobulin A (IgA), the serum level of which was found to decrease 2 days after burns (< 0.001) [15].Consequently, it leads to increased elimination due to less protein binding.Our data was compatible with this finding by lower albumin level and higher vancomycin clearance in burn group.
Besides higher clearance in the burn group, the V d is also higher in the burn group.This could be attributed to more aggressive resuscitation in the burn group than that in the non-burn group.According to Parkland formula, the total amount of resuscitation fluids for burn group was around 14.4 liter (TBSA (%) × body weight × 4 mL), while only around 2 liter (30 mL/kg) was given for non-burn group based on the recommendation of fluid resuscitation in sepsis patients [16,17].This also further explains why the C trough was not significantly higher in the burn group than that in the non-burn group although larger dose was administered to burn group.
Our results have shown that C trough was not correlated well to AUC 24h in burns patients than that in control patients.The result is compatible to the statements by Michael N. Neely who said that the traditional approach to therapeutic drug monitoring (TDM) by comparing the      C trough to a predefined range is a very poor surrogate for estimation of the AUC and overall vancomycin exposure [9].He emphasized that C trough will underestimate the true AUC 24h by about 25%.However, on the other hand, our study also has shown that C trough was significantly correlated with AUC 24h in the non-burn patient group.These findings are compatible to IDSA guidelines, which stated that C trough is a tool to monitor the therapeutic level and toxicity of vancomycin due to its good correlation to AUC 24h in the general population [1].In fact, the association between C trough and AUC 24h may be explained by the pathophysiological changes in burns.Following the ebb phase of early severe burn injury, more aggressive fluid resuscitation than that in control group results in higher than normal cardiac output and greater blood perfusion to the liver and kidneys, consequently increases vancomycin clearance and V d [2].Besides, there is tremendous interpatient or intrapatient variability, especially in burns of different stages.These factors may explain why C trough may not be an appropriate reference to assess the optimal therapeutic vancomycin level in burns.
Some studies showed that the correlation between C trough and the actual AUC 24h were subjected to individual variations such as dosing interval and infusion time [9,10].Again, these findings were against the approach to use C trough alone as a reference.C trough would be proportional to AUC 24h only under the situation of fixing dosing interval and infusion time.In burn patients, higher dose of vancomycin is needed and this makes fixing infusion time difficult due to the concern of red man syndrome [18].This further ensures the necessity of estimate of AUC 24h by at least 2 points of serum concentration, instead of C trough alone.
As C trough does not correlate to AUC 24h well in severe burns patients, two points of serum concentration were suggested in our study for the mathematical derivation of PK parameters and AUC 24h , which are beneficial in vancomycin serum level monitoring and dose adjustment in severe burns [19].Because of increasing CL and V d in severe burn patients, the optimal dose of vancomycin in this population would be higher than that in general population.If vancomycin MIC of MRSA was 1 mg/L, 5000 mg of vancomycin daily was needed to have the better chance to achieve the target of AUC 24h /MIC>400 in this study.Because of vancomycin MIC creep in MRSA and a high percentage of vancomycin MIC being 1 mg/L, the initial vancomycin dosing of 5000 mg should be given in severe burn patients with normal renal function (CL Cr> 90 mL/min) before confirming MIC data [20].However, due to wide individual variations among burn patients, closed serum level monitoring is still highly recommended.As the half-life of vancomycin (4.8 ± 2.5 hours) is comparatively short in burn patients, serum concentration can be checked 1 to 2 days after initiation of vancomycin to ensure efficacy and safety.
Our results have confirmed that large burn size and initiation of vancomycin infusion within 2 weeks of burn in our study population have yielded a high value of CL and V d .The results are compatible to the statement that burn size and timing of vancomycin infusion affect PK [5,21].As vancomycin is mainly excreted by kidneys, previous study has shown that estimated CL Cr is a good predictor of vancomycin clearance [22].However, this is not true in our study, as forming an initial dose nomogram based on CL Cr is not possible (Figure 1).These findings are compatible to other study results stating that creatinine-based formula is a less reliable tool for therapeutic drug monitoring in severe burn patients.The complicated burn pathophysiology has shown that cystatin C-based formula, Cockcroft and Gault methods, Modification of Diet in Renal Disease (MDRD) equations and Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation are imprecise and inaccurate for renal function monitoring [23,24].Besides dehydration due to the tremendous loss of fluid, the suppressed renal function by toxic substances or the decreased renal flow, the decreased serum creatinine production due to the loss of muscle mass, the increased glomerular filtration rate due to the increased cardiac output during the hypermetabolic state, and the disturbed liver metabolism have significantly affect the reliability of CL Cr measurement as a tool for antibiotic dosing.In fact, the glomerular filtration rate is overestimated by creatinine clearance in renal failure due to tubular creatinine secretion [25].In short, the use of creatinine-based formula which is only suitable to patients with stable renal function is inappropriate in severe burn patients.However, urine collection for creatinine clearance remains the clinically gold standard for antibiotic dosing in these patients.
The main drawback of the study is quite small sample size, besides the inherent limitations of retrospective studies.The differences of vancomycin V d between groups may be significant if the sample size is large, and multi-centre prospective study is thus indicated.Also, further studies are needed to confirm the dose recommended in this study.

Conclusion
Our results have proved that C trough alone was not an appropriate reference for monitoring the therapeutic and toxic effect of vancomycin in severe burn patients due to its poor correlation with AUC 24h , in which two points of serum concentration are required.Creatinine clearancebased dosage formula is not feasible due to its lack of correlation with vancomycin clearance in burn patients.We suggest a daily dose of at least 5000 mg to attain an optimal therapeutic value of AUC 24h /MIC ≧ 400 if vancomycin MIC was 1 mg/L of MRSA in severe burn patients with normal renal function.
The final model was assessed by visual inspection of the following diagnostic plots: (a) observed concentrations vs. individual predicted concentrations; (b) observed concentrations vs. predicted concentrations; (c) conditional weighted residuals (CWRES) vs. time and (d) absolute individual weighted residuals (IWRES) vs. individual predictions.A bootstrap resampling technique with 200 replications was conducted to evaluate the stability of the final model using Wings for NONMEM (WFN; Version 6.1, University of Auckland, New Zealand

Figure 2A :
Figure 2A: Area under the curve vs. vancomycin trough concentration: Burn group.

Figure 4 :
Figure 4: Probability of target attainment of AUC 24h /MIC>400 for 5000 simulated burn subjects following different daily doses of vancomycin.

Table 1 :
Demographic data in the burn and matched control groups.
a B: Burn group; C: Matched cohort, b M: Male; F: Female, c TBSA: Total body surface area of second and third degree burn # obese * p<0.05 compared to control group

Table 2 :
Vancomycin data in the burn and matched control groups.