JBUR-4077; No. of Pages 8
Amikacin population pharmacokinetics amongpaediatric burn patients
Catherine M.T. Sherwin a,*, Stephanie Wead b, Chris Stockmann a,Daniel Healy b,d, Michael G. Spigarelli a, Alice Neely c,d, Richard Kagan c,d
aDivision of Clinical Pharmacology, Department of Paediatrics, University of Utah School of Medicine, Salt Lake City,
Utah, United Statesb James L. Winkle College of Pharmacy, University of Cincinnati, Cincinnati, Ohio, United StatescDepartment of Surgery, University of Cincinnati, Cincinnati, Ohio, United StatesdThe Shriners Hospitals for Children1, Cincinnati, Ohio, United States
b u r n s x x x ( 2 0 1 3 ) x x x – x x x
a r t i c l e i n f o
Article history:
Accepted 14 June 2013
Keywords:
Amikacin
Pharmacokinetics
Paediatric burns
a b s t r a c t
Introduction: The objectives of this study were to (1) determine the pharmacokinetics of
amikacin among children with severe burn and (2) identify influential covariates.
Methods: Population-based pharmacokinetic modelling was performed in NONMEM 7.2 for
hospitalized children who received amikacin at 10–20 mg/kg divided two, three, or four
times per day as part of early empiric treatment of presumed burn-related sepsis.
Results: The analysis included data from 70 patients (6 months to 17 years) with 282
amikacin serum concentrations. Amikacin’s mean Cmax was 33.2 � 9.4 mg/mL and the mean
Cmin was 3.8 � 4.6 mg/mL. The final covariate model estimated clearance as 5.98 L/h/70 kg
(4.97–6.99, 95% CI), the volume of distribution in the central compartment as 16.7 L/70 kg
(14.0–19.4, 95% CI), the volume of distribution in the peripheral compartment as 40.1 L/70 kg
(15.0–80.4, 95% CI), and the inter-compartmental clearance as 3.38 L/h/70 kg (2.44–4.32, 95% CI).
In multivariate analyses, current weight (P < 0.001) was a significant covariate, while age, sex,
height, serum creatinine, C-reactive protein, platelet count, the extent and type of burn, and
concomitant vancomycin administration did not influence amikacin pharmacokinetics.
Discussion: Children with burn featured elevated amikacin clearance when compared to
healthy adult volunteers. However, peak amikacin concentrations are comparable to those
attained in other critically-ill children, suggesting that elevated amikacin clearance may not
result in sub-therapeutic antibacterial effects. In this study, we found that amikacin dis-
plays two-compartment pharmacokinetics, with weight exerting a strong effect upon
amikacin clearance. Further pharmacodynamic studies are needed to establish the optimal
dosing regimen for amikacin in paediatric burn patients.
# 2013 Elsevier Ltd and ISBI. All rights reserved.
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/burns
1. Introduction
The aminoglycoside antibiotic amikacin is a mainstay of
treatment for gram-negative sepsis among critically-ill burn
* Corresponding author at: University of Utah Health Sciences Center, 2United States. Tel.: +1 801 587 7404; fax: +1 801 585 9410.
E-mail address: [email protected] (Catherine M.T.
Please cite this article in press as: Sherwin CMT, et al. Amikacin populathttp://dx.doi.org/10.1016/j.burns.2013.06.015
0305-4179/$36.00 # 2013 Elsevier Ltd and ISBI. All rights reserved.http://dx.doi.org/10.1016/j.burns.2013.06.015
patients [1,2]. The pathophysiology of burn can result in
substantially altered aminoglycoside pharmacokinetics, in-
cluding heightened renal clearance, increased volume of
distribution, and altered protein binding [3–6]. As a conse-
quence of these pharmacokinetic changes, previous reports
95 Chipeta Way, Clinical Pharmacology, Salt Lake City, Utah 84108,
Sherwin).
ion pharmacokinetics among paediatric burn patients. Burns (2013),
b u r n s x x x ( 2 0 1 3 ) x x x – x x x2
JBUR-4077; No. of Pages 8
have recommended higher aminoglycoside doses and/or more
frequent administration [1,7]. However, aminoglycosides may
adversely affect auditory, vestibular, and renal function,
adding further complexity to dosing in this patient population
[8]. Investigation of amikacin pharmacokinetic parameters is
essential for defining factors that contribute to variability in
drug disposition and in determining an optimal dosing
regimen for paediatric burn patients.
Amikacin features concentration-dependent bactericidal
activity and a substantial post-antibiotic effect [9–11]. Howev-
er, similar to other aminoglycoside antibiotics, adaptive
resistance to amikacin is enhanced by continued presence
of the drug [12]. Previous reports have suggested that the
development of adaptive resistance may be minimized during
once-daily dosing [13–15]. Additionally, amikacin toxicity may
occur with sustained drug concentrations [16]. Nephrotoxicity
occurs as a consequence of amikacin accumulation in the
proximal renal tubules [17]. However, Williams et al. demon-
strated that once proximal tubular cells have been saturated
with amikacin, increasing concentrations are not likely to
result in increased intracellular amikacin accumulation
[18,19]. Amikacin use has also been associated with loss of
cochlear and vestibular hair cells, leading to hearing loss and
disequilibrium [20]. The risk of ototoxicity has been strongly
associated with high trough concentrations [21]. In an attempt
to enhance efficacy and reduce the risk of nephrotoxicity and
ototoxicity once-daily dosing of amikacin has become com-
mon practice among non-burn patients [22,23]. However,
limited pharmacologic and clinical evidence exists to aid in
the determination of an optimal, individualized amikacin
dosing regimen for severely-burned paediatric patients.
The primary objective of this study was to evaluate the
pharmacokinetic parameters of amikacin among critically-ill
children with severe burn. As a secondary aim, covariates
which influence amikacin pharmacokinetic parameters were
evaluated. We hypothesized that burned children would more
rapidly clear amikacin than healthy volunteers, which could
potentially result in sub-therapeutic antibacterial activity.
Ultimately, improved understanding of amikacin pharmaco-
kinetics in this population offers the opportunity to develop
optimal dosing regimens and aid in the design of future
studies conducted among paediatric burn patients.
2. Methods
2.1. Subjects
This study involved 73 paediatric burn patients who were
hospitalized in the dedicated burn unit at the Cincinnati
Shriners Hospital for Children, Cincinnati, Ohio. All patients
received amikacin as part of an empiric regimen with
piperacillin/tazobactam and vancomycin for presumed or
proven burn wound sepsis. Patient demographics, including:
age, sex, weight, height, per cent total body surface area burn,
and serum creatinine were recorded.
This study was reviewed and approved by the University of
Cincinnati Institutional Review Board. Parental permission
and informed assent (when appropriate) were obtained prior
to the performance of any study-related procedures.
Please cite this article in press as: Sherwin CMT, et al. Amikacin populathttp://dx.doi.org/10.1016/j.burns.2013.06.015
2.2. Drug administration
Amikacin was infused over 30 min using a syringe pump at a
daily dose of 10–20 mg/kg divided two (13% of patients), three
(45%), or four (42%) times. Dosing adjustments (e.g., decreased/
increased dose or shorten/lengthen dosing interval) were
made to target peak concentrations from 25 to 30 mg/mL and
trough concentrations of 4–8 mg/mL.
2.3. Sample collection
Blood samples were obtained for routine therapeutic drug
monitoring for medical care. Samples were drawn within
30 min before the dose (trough concentration) and 1 h after the
end of the intravenous infusion (peak concentration). The
duration of treatment was determined on the basis of the
child’s clinical status and results of microbiologic testing.
2.4. Analytical assay
Amikacin serum concentrations were measured using an
automated fluorescence polarization immunoassay (Abbott
TDx, Abbott Park, IL) [24]. The assay was linear from 0.8–
50.0 mg/mL with reported within-day and between-day coeffi-
cients of variation <7.6% for low (5 mg/mL), medium (15 mg/
mL), and high (30 mg/mL) quality control specimens that were
run with each batch of patient unknowns.
2.5. Pharmacokinetic analysis
For a detailed description of the pharmacokinetic modelling
methods please refer to the Appendix. Briefly, amikacin
pharmacokinetics were evaluated in NONMEM 7.2 (non-linear
mixed effects modelling; ICON Development Solutions, Ellicott
City, MD). Non-linear mixed effects models were used to
describe the amikacin concentration-time response for this
population of burned children. Population estimates of
amikacin clearance (CL) and volume of distribution (Vd) were
calculated based upon the dose and time of administration for
each patient.
Model variability and random effects were classified as
belonging to one of two types of error: (1) between-subject
variability (BSV) and (2) residual unexplained variability (RUV).
BSV is the inherent variability between individual subjects.
RUV reflects the difference between the model prediction for
the individual and the measured concentration. This includes
errors in the time of measurement, errors in the drug dose,
error in the assay, etc. [25]. Estimates of these errors were
calculated using additive, proportional, and combined error
models.
Patient age, sex, weight, height, per cent total body surface
area burn, and serum creatinine were evaluated for their effect
upon amikacin pharmacokinetic parameters. Initially, rela-
tionships between potential covariates and amikacin phar-
macokinetics were assessed in generalized additive models.
Further covariate selection was performed by the stepwise
addition of covariates that significantly improved the model’s
fit (P < 0.05). Covariates were then eliminated in a backward
stepwise procedure if they did not result in a highly significant
improvement to the model’s fit (P < 0.01). The final covariate
ion pharmacokinetics among paediatric burn patients. Burns (2013),
Table 1 – Demographic characteristics among severely-burned children who received amikacin for proven orpresumed gram-negative infections.
Characteristic Number (%)(n = 70)
Age (years)
Median 4.5
Range 0.6–17
Sex
Boys 45 (64)
Girls 25 (36)
Race/Ethnicity
White 37 (53)
Black 13 (19)
Hispanic 16 (23)
Asian/Pacific Islander 2 (3)
Other 2 (3)
Weight (kg)
Median 20
Range 8–90
Concomitant vancomycin use
Yes 62 (89)
No 8 (11)
Type of burn
Flame 47 (67)
Scald 18 (26)
Electrical 5 (7)
Inhalation injury
Yes 13 (19)
No 57 (81)
Per cent total body surface area burned
Median 43
Range 11–98
Fig. 1 – Diagnostic plot comparing weight versus age.
(regression line shown as a dashed line.).
b u r n s x x x ( 2 0 1 3 ) x x x – x x x 3
JBUR-4077; No. of Pages 8
model was used to estimate amikacin’s pharmacokinetic
parameters in this population of severely burned children.
3. Results
3.1. Subjects and pharmacokinetics
Three patients were excluded from the pharmacokinetic
analysis due to missing data. From the remaining 70 patients,
there were 282 amikacin concentrations within the dataset.
Patients were dosed at 10 to 20 mg/kg/day divided 2, 3, or 4
times (actual dosages ranged from 4.9 to 22.3 mg/kg; mean
16.4 � 3.9 mg/kg). Amikacin’s mean peak concentration (Cmax)
was determined to be 33.2 � 9.4 mg/mL. The mean trough
concentration (Cmin) was 3.8 � 4.6 mg/mL.
The median age of the children included in this study was
4.5 (range: 0.6–17) years. A majority of subjects were boys (64%)
and white (53%). The median body weight of the patients was
20 (range: 8–90) kg (Table 1). Flame injuries were the most
common (67%) type of burn. A minority of children (19%)
suffered inhalation injuries. The median per cent total body
surface area burned was 43% (range: 11–98%).
3.2. Population pharmacokinetic models
Multiple structural models were explored to determine the
model that best fit the amikacin concentration data.
Please cite this article in press as: Sherwin CMT, et al. Amikacin populathttp://dx.doi.org/10.1016/j.burns.2013.06.015
One-compartment and two-compartment structural models
were assessed. A combined residual error model resulted in
the greatest improvement in the model fit.
One-compartment models with first-order elimination
were explored. However, a two-compartment combined error
model better described the disposition of amikacin in these
severely-burned children. Using a two-compartment model,
the base model estimated CL at 2.48 L/h (2.12–2.84 95% CI); the
volume of distribution in the central compartment (VC) was
2.71 L (2.05–3.37 95% CI) and the volume of distribution in the
peripheral compartment (VP) was 8.53 L (3.13–13.9 95% CI).
The inter-compartmental clearance Q was estimated as
1.13 L/h (0.79–1.47 95% CI). Estimates of the BSV in CL and
VC were 52.5% and 42.9%, respectively. The BSV in Q was fixed
at 10%. The RUV coefficient of variation was estimated at
23.9%.
3.3. Covariate models
Using univariate analyses, weight (P < 0.001), age (P < 0.05),
and concomitant vancomycin administration (P < 0.05) were
identified as having a significant influence upon amikacin
pharmacokinetics. However, in multivariate analyses ac-
counting for weight, age and vancomycin use did not
significantly influence the volume of distribution (Fig. 1).
The final covariate model was chosen as it produced the
best model fit, reduced the BSV, and decreased the RUV (Figs. 2
and 3). The parameter estimates derived from the final
covariate model are featured in Table 2.
3.4. Model evaluation
Diagnostic plots were generated for observed amikacin
concentrations versus population predicted and individual
predicted values (Figs. 2 and 3). Plots of residuals and
conditional weighted residuals versus time after dose and
population predicted amikacin concentrations were also
examined (Fig. 4).
Mean estimates from the 1000 bootstrap runs were similar
to the population estimates derived from the final covariate
model. Bootstraps were successfully generated 100% of the
time. The final covariate model generated reasonably stable
ion pharmacokinetics among paediatric burn patients. Burns (2013),
Fig. 2 – Diagnostic plots of the observed versus population predicted amikacin concentrations for the (A) base two-
compartment model and(B) final two-compartment covariate model. (Regression lines are shown as dashed lines.).
b u r n s x x x ( 2 0 1 3 ) x x x – x x x4
JBUR-4077; No. of Pages 8
and accurate estimates of the fixed and random effects.
Simulations from the observed amikacin data are presented in
Fig. 5 using a visual predictive check (VPC), with the median
simulated value compared to the 5th, 10th, 90th, and 95th
quantiles. Of 58,590 simulated observations 94.6% fell within
the 90% confidence interval, demonstrating model stability
and reasonable agreement between the observed and simu-
lated amikacin concentration data.
4. Discussion
This study examined amikacin’s disposition in paediatric burn
patients and identified patient characteristics, specifically
weight, that influence the pharmacokinetics of amikacin.
Although the pharmacokinetics of amikacin have been well
described among healthy adult volunteers [26–28], few studies
have examined amikacin’s disposition in patients with severe
burns [7,29]. Moreover, several previous studies have assumed
a one-compartment model [29,30], although this study found
that a two-compartment model was superior for patients with
burns. Peak serum concentrations were comparable to
previous reports from paediatric burn patients and other
critically-ill children (29.4 � 4.2 mg/mL reported by Zaske et al.
vs. 33.2 � 9.4 reported here) [1,31]. However, the original Zaske
study included data from 3 children and 7 adults, whereas this
study examined the pharmacokinetics of amikacin in children
Fig. 3 – Diagnostic plots of the observed versus individual pred
compartment model and (B) final two-compartment covariate m
Please cite this article in press as: Sherwin CMT, et al. Amikacin populathttp://dx.doi.org/10.1016/j.burns.2013.06.015
only. Bressolle et al. evaluated the population pharmacoki-
netics of amikacin in a population of critically-ill children
without burns injuries; however, their population model could
not be used to establish a constant CL value, as CL was strongly
influenced by demographic and pathological characteristics of
their study cohort [32]. In this study, we applied allometric
scaling and estimated CL as 5.98 L/h/70 kg, which is compara-
ble to the range of CL values reported by Bressolle et al.
Additionally, this study corroborates earlier reports by
identifying current body weight as a significant covariate that
influences amikacin’s volume of distribution [32–34].
Effective and safe aminoglycoside dosing among children
with severe burn is challenging due to their shorter half-life,
faster CL, and larger volume of distribution when compared
with healthy volunteers [35,36]. In a non-compartmental
analysis of amikacin’s pharmacokinetics among 38 paediatric
burn patients performed over 20 years ago at our institution,
Kopcha et al. found that the volume of distribution varied
widely and decreased with increasing age [7]. On the basis of
these findings, the authors concluded that doses above the
manufacturer’s recommendations are required to achieve
therapeutic concentrations in children with severe burns. The
present study applied a two-compartment model and found
that VP varied substantially, while VC was relatively stable.
Moreover, in multivariate analyses age did not significantly
influence amikacin pharmacokinetics; however, current
weight was correlated with the volume of distribution in
icted amikacin concentrations for the (A) base two-
odel. (Regression lines are shown as dashed lines.).
ion pharmacokinetics among paediatric burn patients. Burns (2013),
Table 2 – Amikacin pharmacokinetic parameter estimates and bootstrap estimates from the final two-compartmentcovariate model.
Parameters Parameterestimates
% RSEa % CVb 95% CIc Bootstrapmean (n = 1000)
95% CIc
Pharmacokinetic parameters
Clearance (CL), L/h/70 kg 5.98 8.6 – 4.97–6.99 5.57 4.0–6.18
Volume of distribution in the
central compartment (VC), L/70 kg
16.7 8.14 – 14.0–19.4 15.8 2.58–19.1
Intercompartmental clearance (Q), L/h/70 kg 3.38 14.2 – 2.44–4.32 3.96 2.95–6.56
Volume of distribution in the peripheral
compartment (VP), L/70 kg
40.1 34.3 – 15.8–80.4 57.1 8.1–101.0
Between subject variability (BSV)
BSV (v)–Clearance (CL) 0.0874 26.4 29.6 0.0421–0.133 0.087 0.003–0.12
BSV (v)–Volume of distribution in the
central compartment (VC)
0.06 9.6 24.2 �0.00442 to 0.121d 0.26 0.00002–2.2
BSV (v)–Intercompartmental clearance
(Q), Fixed
0.01 – – – 0.003 –
Residual unexplained variability (RUV)
RUV (s)–Standard deviation (mg/mL) 1.81 45.14 1.35 0.211–3.41 1.183 0.29–4.34
RUV (s)–Coefficient of variation (%) 0.0385 33.24 19.6 0.0134–0.0636 0.049 0.056–0.09
a Per cent root mean square error.b Per cent coefficient of variance.c Ninety-five per cent confidence interval.d 95% CI includes zero.
b u r n s x x x ( 2 0 1 3 ) x x x – x x x 5
JBUR-4077; No. of Pages 8
the central compartment. Previous studies in neonates have
also identified current weight as an important determinant of
amikacin volume of distribution [34].
Although increased amikacin clearance has been frequent-
ly reported among burn patients [30], the pharmacokinetic
mechanism for this is not completely understood. As
amikacin is almost exclusively eliminated via glomerular
filtration, an increase in the glomerular filtration rate (as
estimated by creatinine clearance) may contribute to height-
ened amikacin clearance [37]. Conil et al. identified a strong
correlation between amikacin elimination and creatinine
clearance among a cohort of 38 adults with severe burn [30].
However, additional studies have also suggested that tubular
secretion may play a role in the elimination of amikacin
[38,39]. In aggregate, these studies suggest that amikacin’s
heightened clearance in burn patients may reflect some
combination of both increased glomerular filtration rate and
increased tubular secretion.
Fig. 4 – Diagnostic plots of the conditional weighted residual vers
base two-compartment model and (B) final two-compartment c
lines.).
Please cite this article in press as: Sherwin CMT, et al. Amikacin populathttp://dx.doi.org/10.1016/j.burns.2013.06.015
Adult patients with severe burn have also been reported to
feature altered amikacin pharmacokinetics [30,38]. Conil et al.
found that conventional once-daily dosing did not consistent-
ly achieve peak concentration targets that have been shown to
be predictive of clinical efficacy [30]. Rong-hua et al. evaluated
amikacin pharmacokinetics in the sub-eschar fluid of patients
with early stage burn [38]. They reported half-lives of amikacin
in the sub-eschar fluid that were 25–40-fold longer than those
found in healthy volunteers. Additionally, amikacin concen-
trations remained above the minimum inhibitory concentra-
tion (MIC) of many common pathogenic bacteria at 24 h after a
single dose infusion. This suggests that amikacin can easily
diffuse into the sub-eschar and intra-eschar tissue and may
provide effective inhibitory concentrations for a sustained
period of time among severely burned patients, despite
seemingly low serum concentrations.
Elevated amikacin clearance, as seen in burn, is often
accompanied by renal failure in critically-ill patients. Akers
us population predicted amikacin concentrations for the (A)
ovariate model. (Regression lines are shown as dashed
ion pharmacokinetics among paediatric burn patients. Burns (2013),
Fig. 5 – Visual predictive check for final covariate model,
amikacin observed data compared with the 95th, 50th and
5th percentiles for 100 simulated data sets. Observed data
for 36 h. Comparison of median (dashed line) and 5th–95th
percentile interval (solid black lines). 50th quantiles ( );
10th–90th quantiles ( ); 5th–95th quantiles ( ).
b u r n s x x x ( 2 0 1 3 ) x x x – x x x6
JBUR-4077; No. of Pages 8
et al. evaluated amikacin pharmacokinetics among 60 burn
patients, 12 of whom were treated with continuous venove-
nous hemofiltration [29]. Among all patients, only 8.5%
achieved the preferred amikacin pharmacodynamic target
of a maximum concentration to MIC ratio � 10. Although
mortality and burn size were higher among patients receiving
renal replacement therapy, amikacin clearance was not
significantly affected. The authors suggest that higher
amikacin MICs and suboptimal dosing are the primary
determinants of low pharmacodynamic target attainment,
as opposed to increased CL related to renal replacement
therapy. In another pharmacokinetic study conducted among
patients receiving high-dose chemotherapy, Davis et al.
observed higher amikacin volume of distribution, CL and
elimination half-life when compared to healthy volunteers
[40].
This study is subject to several limitations. These data were
collected during routine therapeutic drug monitoring and a
limited number of observations were available for each
patient. Additionally, this study was not designed to correlate
amikacin pharmacokinetics with clinical efficacy, although
several earlier studies have suggested that high peak
concentrations and the speed with which peak concentrations
are achieved are positively correlated with improved clinical
outcomes [41–43]. More recent studies have clearly established
the influence of appropriate early antibiotic therapy on
survival [44,45]. This includes appropriate serum concentra-
tions with an antibiotic that provides targeted coverage for the
pathogenic organism.
In conclusion, children with burn feature elevated amika-
cin clearance when compared to healthy adult volunteers.
However, peak amikacin concentrations are comparable to
those attained in other critically-ill children, suggesting that
elevated amikacin clearance may not result in sub-therapeutic
Please cite this article in press as: Sherwin CMT, et al. Amikacin populathttp://dx.doi.org/10.1016/j.burns.2013.06.015
antibacterial effects. In this study, we found that amikacin
displays two-compartment pharmacokinetics, with weight
exerting a strong effect upon amikacin clearance. The extent
of burn, serum creatinine, and inflammatory biomarkers did
not significantly influence amikacin pharmacokinetics. Fur-
ther research is warranted to determine if dosage adjustments
optimize the safety and efficacy of amikacin for paediatric
burn patients.
5. Conflicts of interest
All authors have completed and submitted the ICMJE Form for
Disclosure of Potential Conflicts of Interest. Dr. Kagan serves
as a consultant for Johnson & Bell, Ltd. and Rosselot Law Firm
and has grants pending from the Shriners Hospitals for
Children. The authors declare: [DH, RK, AN, SW] had support
from [Shriners Hospitals for Children] for the submitted work;
no financial relationships with any organisations that might
have an interest in the submitted work in the previous 3 years;
no other relationships or activities that could appear to have
influenced the submitted.
Funding
Shriners Hospitals for Children (grant #70011, DPH).
Role of the sponsor
The funding sponsor had no part in the design and conduct of
the study; collection, management, analysis and interpreta-
tion of the data; and preparation, review, or approval of the
manuscript. The sponsor had no access to the data and did not
perform any of the study analyses.
Acknowledgements
We would like to acknowledge Mary Rieman, RN clinical
research coordinator and the other clinical research staff as
well as the Departments of Microbiology and Laboratory
Medicine at SHC-Cincinnati for facilitating data collection. We
would also like to thank our summer student Patrick Muvunyi
for cleaning and formatting the dataset.
Appendix
Population pharmacokinetic analysis
Amikacin’s pharmacokinetic parameters were derived using
NONMEM 7.2 (non-linear mixed effects modelling; ICON
Development Solutions, Ellicott City, MD). One- and two-
compartment structural models were fitted to the data. The
one-compartment model enabled the estimation of amikacin
clearance (CL) and the volume of distribution (Vd). The two-
compartment model was parameterized to give estimates of
amikacin CL, central compartment volume of distribution (VC),
ion pharmacokinetics among paediatric burn patients. Burns (2013),
b u r n s x x x ( 2 0 1 3 ) x x x – x x x 7
JBUR-4077; No. of Pages 8
peripheral compartment volume of distribution (VP), and
inter-compartmental clearance (Q).
Between subject variability (BSV) was assumed to be log-
normally distributed and was assessed using an exponential
equation of the form:
Pi ¼ upop � expðhrÞ; (1)
where Pi is the value of the pharmacokinetic parameters for
the ith individual, upop is the population mean for P, and h
represents the between subject random effect with a mean of
zero and a variance of v2.
During model development, residual unexplained variabil-
ity (RUV) was evaluated using additive, proportional, and
combined error models. The combined error model followed
the form of:
Yi j ¼ Ymi jð1 þ ei jÞ þ ei j; (2)
where Yij is the observed concentration for the ith individual at
time j, Ymij is the model prediction, and eij is a normally-
distributed random error with a mean of zero and a variance
of s2.
Covariate analysis
Plots were generated to perform exploratory analyses and
assess the relationships between parameter estimates and
potential covariates. Age, weight, height, C-reactive protein
(CRP), platelet count, serum creatinine, and concomitant
vancomycin administration were included in the covariate
analysis. All pharmacokinetic parameters were scaled accord-
ing to standard allometric equations (Eq. (3)) in which
parameter estimates determined for this paediatric popula-
tion were standardized to values reported for a typical, 70 kg
adult.
CLi ¼ CLpop �BW70
� �u;allo !
� expðhCLÞ; (3)
where CLi is the individual clearance in the ith individual,
CLpop is the estimate of the population clearance, hCL is the
random between subject variability, u,allo is a fixed allometric
power parameter that was assigned a value of 0.75 to describe
the systematic dependence of clearance on individual body
weight and a value of 1 for the volume of distribution, and BW
is the body weight of the ith individual.
Additionally, the type of burn (e.g., flame, scald, and
electrical burns), the per cent of total body surface area
burned, and the per cent full-thickness and partial-thickness
burns were included as potential covariates. Each covariate
was assessed separately in univariate analyses. Covariates
which were significant in univariate analyses were then
included in multivariate analyses through a backward and
forward stepwise selection process.
Models were compared by examining residual plots, the
precision of parameter estimates, measures of variability, and
the objective function value (OFV). Residual plots were used to
discriminate between different weightings and the F-test was
used to compare the weighted sum of squared residuals
(WRSS) among identically weighted one- and two-compart-
ment models. Model fit was based on minimization of the OFV.
Please cite this article in press as: Sherwin CMT, et al. Amikacin populathttp://dx.doi.org/10.1016/j.burns.2013.06.015
A reduction of more than 3.84 (-2 log likelihood difference) was
considered statistically significant with P < 0.05 and one
degree of freedom.
Model evaluation
To assess uncertainty in parameter estimates nonparametric
bootstrapping techniques were applied to the final pharma-
cokinetic models [46]. PDx-Pop was used to derive 1000
bootstrap runs by randomly sampling with replacement from
the original dataset. Standard errors were assessed for both
the estimated population parameters and random effects
error models. Bootstrap techniques, goodness-of-fit plots, and
visual predictive checks were used to evaluate model fit.
r e f e r e n c e s
[1] Zaske DE, Sawchuk RJ, Strate RG. The necessity of increaseddoses of amikacin in burn patients. Surgery 1978;84:603–8.
[2] Nguyen TT, Gilpin DA, Meyer NA, Herndon DN. Currenttreatment of severely burned patients. Ann Sur1996;223:14–25.
[3] Weinbren MJ. Pharmacokinetics of antibiotics in burnpatients. J Antimicrob Chemother 1999;44:319–27.
[4] Smeltzer BD, Schwartzman MS, Bertino Jr JS. Amikacinpharmacokinetics during continuous ambulatoryperitoneal dialysis. Antimicrob Agents Chemother1988;32:236–40.
[5] Boucher BA, Kuhl DA, Hickerson WL. Pharmacokinetics ofsystemically administered antibiotics in patients withthermal injury. Clinical Infect Dis 1992;14:458–63.
[6] Jaehde U, Sorgel F. Clinical pharmacokinetics in patientswith burns. Clin Pharmacokinet 1995;29:15–28.
[7] Kopcha RG, Fant WK, Warden GD. Increased dosingrequirements for amikacin in burned children. J AntimicrobChemother 1991;28:747–52.
[8] Ali MZ, Goetz MB. A meta-analysis of the relative efficacyand toxicity of single daily dosing versus multiple dailydosing of aminoglycosides. Clin Infect Dis 1997;24:796–809.
[9] Garraffo R, Drugeon HB, Dellamonica P, Bernard E, LapalusP. Determination of optimal dosage regimen for amikacinin healthy volunteers by study of pharmacokinetics andbactericidal activity. Antimicrob Agents Chemother1990;34:614–21.
[10] Blanchet B, Jullien V, Vinsonneau C, Tod M. Influence ofburns on pharmacokinetics and pharmacodynamics ofdrugs used in the care of burn patients. Clin Pharmacokin2008;47:635–54.
[11] Burgess DS. Use of pharmacokinetics andpharmacodynamics to optimize antimicrobial treatment ofPseudomonas aeruginosa infections. Clin Infect Dis2005;40(Suppl. 2):S99–104.
[12] Daikos GL, Jackson GG, Lolans VT, Livermore DM. Adaptiveresistance to aminoglycoside antibiotics from first-exposure down-regulation. J Infect Dis 1990;162:414–20.
[13] Fisman DN, Kaye KM. Once-daily dosing of aminoglycosideantibiotics. Infect Dis Clin N Am 2000;14:475–87.
[14] Karlowsky JA, Zelenitsky SA, Zhanel GG. Aminoglycosideadaptive resistance. Pharmacotherapy 1997;17:549–55.
[15] Daikos GL, Lolans VT, Jackson GG. First-exposure adaptiveresistance to aminoglycoside antibiotics in vivo withmeaning for optimal clinical use. Antimicrob AgentsChemother 1991;35:117–23.
ion pharmacokinetics among paediatric burn patients. Burns (2013),
b u r n s x x x ( 2 0 1 3 ) x x x – x x x8
JBUR-4077; No. of Pages 8
[16] Kahlmeter G, Dahlager JI. Aminoglycoside toxicity – areview of clinical studies published between 1975 and 1982.J Antimicrob Chemother 1984;13 (Suppl. A):9–22.
[17] De Broe ME, Verbist L, Verpooten GA. Influence of dosageschedule on renal cortical accumulation of amikacin andtobramycin in man. J Antimicrob Chemother 1991;27(Suppl. C):41–7.
[18] Contrepois A, Brion N, Garaud JJ, Faurisson F, Delatour F,Levy JC, et al. Renal disposition of gentamicin, dibekacin,tobramycin, netilmicin, and amikacin in humans.Antimicrob Agents Chemother 1985;27:520–4.
[19] Williams PD, Bennett DB, Gleason CR, Hottendorf GH.Correlation between renal membrane binding andnephrotoxicity of aminoglycosides. Antimicrob AgentsChemother 1987;31:570–4.
[20] Chen Y, Huang WG, Zha DJ, Qiu JH, Wang JL, Sha SH, et al.Aspirin attenuates gentamicin ototoxicity: from thelaboratory to the clinic. Hear Res 2007;226:178–82.
[21] Black RE, Lau WK, Weinstein RJ, Young LS, Hewitt WL.Ototoxicity of amikacin. Antimicrob Agents Chemother1976;9:956–61.
[22] Barclay ML, Kirkpatrick CM, Begg EJ. Once dailyaminoglycoside therapy. Is it less toxic than multiple dailydoses and how should it be monitored? Clin Pharmacokin1999;36:89–98.
[23] Ferriols-Lisart R, Alos-Alminana M. Effectiveness andsafety of once-daily aminoglycosides: a meta-analysis. AmJ Health Syst Pharm 1996;53:1141–50.
[24] Jolley ME, Stroupe SD, Wang CH, Panas HN, Keegan CL,Schmidt RL, et al. Fluorescence polarization immunoassay.I. Monitoring aminoglycoside antibiotics in serum andplasma. Clin Chem 1981;27:1190–7.
[25] Ette EI, Williams PJ. Population pharmacokinetics I:background, concepts, and models. Ann Pharmacother2004;38:1702–6.
[26] Kirby WM, Clarke JT, Libke RD, Regamey C. Clinicalpharmacology of amikacin and kanamycin. J Infect Dis1976;134 (Suppl.):S312–5.
[27] Van der Auwera P. Pharmacokinetic evaluation of singledaily dose amikacin. J Antimicrob Chemother 1991;27(Suppl. C):63–71.
[28] Walker JM, Wise R, Mitchard M. The pharmacokinetics ofamikacin and gentamicin in volunteers: a comparison ofindividual differences. J Antimicrob Chemother 1979;5:95–9.
[29] Akers KS, Cota JM, Frei CR, Chung KK, Mende K, Murray CK.Once-daily amikacin dosing in burn patients treated withcontinuous venovenous hemofiltration. Antimicrob AgentsChemother 2011;55:4639–42.
[30] Conil JM, Georges B, Breden A, Segonds C, Lavit M, Seguin T,et al. Increased amikacin dosage requirements in burnpatients receiving a once-daily regimen. Int J AntimicrobAgents 2006;28:226–30.
[31] Zaske DE, Strate RG, Kohls PR. Amikacin pharmacokinetics:wide interpatient variation in 98 patients. J Clin Pharmacol1991;31:158–63.
Please cite this article in press as: Sherwin CMT, et al. Amikacin populathttp://dx.doi.org/10.1016/j.burns.2013.06.015
[32] Bressolle F, Gouby A, Martinez JM, Joubert P, Saissi G,Guillaud R, et al. Population pharmacokinetics of amikacinin critically ill patients. Antimicrob Agents Chemother1996;40:1682–9.
[33] Romano S, Fdez de Gatta MM, Calvo MV, Caballero D,Dominguez-Gil A, Lanao JM. Population pharmacokineticsof amikacin in patients with haematological malignancies.J Antimicrob Chemother 1999;44:235–42.
[34] Sherwin CM, Svahn S, Van der Linden A, Broadbent RS,Medlicott NJ, Reith DM. Individualised dosing of amikacinin neonates: a pharmacokinetic/pharmacodynamicanalysis. Eur J Clin Pharmacol 2009;65:705–13.
[35] Sawchuk RJ, Zaske DE. Pharmacokinetics of dosingregimens which utilize multiple intravenous infusions:gentamicin in burn patients. J Pharmacokin Biopharm1976;4:183–95.
[36] Zaske DE, Sawchuk RJ, Gerding DN, Strate RG. Increaseddosage requirements of gentamicin in burn patients. JTrauma 1976;16:824–8.
[37] De Cock RF, Allegaert K, Schreuder MF, Sherwin CM, deHoog M, van den Anker JN, et al. Maturation of theglomerular filtration rate in neonates, as reflected byamikacin clearance. Clin Pharmacokin 2012;51:105–17.
[38] Yang RH, Rong XZ, Hua R, Zhang T. Pharmacokinectics ofvancomycin and amikacin in the subeschar tissue fluid inpatients with severe burn. Burns 2009;35:75–9.
[39] Drozdzik M, Domanski L, Wojcicki J, Gawronska-Szklarz B,Machoy P, Pudlo A. Effect of unilateral nephrectomy on thepharmacokinetics of amikacin in humans. J PharmPharmacol 2002;54:509–14.
[40] Davis RL, Lehmann D, Stidley CA, Neidhart J. Amikacinpharmacokinetics in patients receiving high-dose cancerchemotherapy. Antimicrob Agents Chemother 1991;35:944–7.
[41] Barza M, Lauermann M. Why monitor serum levels ofgentamicin? Clin Pharmacokin 1978;3:202–15.
[42] Hassan E, Ober JD. Predicted and measured aminoglycosidepharmacokinetic parameters in critically ill patients.Antimicrob Agents Chemother 1987;31:1855–8.
[43] Moore RD, Lietman PS, Smith CR. Clinical response toaminoglycoside therapy: importance of the ratio of peakconcentration to minimal inhibitory concentration. J InfectDis 1987;155:93–9.
[44] Gaieski DF, Mikkelsen ME, Band RA, Pines JM, Massone R,Furia FF, et al. Impact of time to antibiotics on survival inpatients with severe sepsis or septic shock in whom earlygoal-directed therapy was initiated in the emergencydepartment. Crit Care Med 2010;38:1045–53.
[45] Kang CI, Kim SH, Kim HB, Park SW, Choe YJ, Oh MD, et al.Pseudomonas aeruginosa bacteremia: risk factors formortality and influence of delayed receipt of effectiveantimicrobial therapy on clinical outcome. Clin Infect Dis2003;37:745–51.
[46] Ette EI. Stability and performance of a populationpharmacokinetic model. J Clin Pharmacol 1997;37:486–95.
ion pharmacokinetics among paediatric burn patients. Burns (2013),