J Antimicrob Chemother 2024; 79: 262–270

https://doi.org/10.1093/jac/dkad368 Advance Access publication 9 December 2023

Effect of albumin substitution on pharmacokinetics of piperacillin/

tazobactam in patients with severe burn injury admitted to the ICU
Beatrix Wulkersdorfer1,2, Felix Bergmann 1,3, Lisa Amann4, Alexandra Fochtmann-Frana3, Valentin Al Jalali 1
,
Elizaveta Kurdina1, Edith Lackner1, Sebastian G. Wicha4, Christoph Dorn5, Bruno Schäfer6, Gerald Ihra6,
Thomas Rath3, Christine Radtke3 and Markus Zeitlinger 1*

1
Medical University of Vienna, Department of Clinical Pharmacology, Währinger Gürtel 18-20, 1090 Vienna, Austria; 2Orthopedic Clinic—
SKA Zicksee, Otto-Pohanka-Platz 1, 7161 St.Andrä am Zicksee, Austria; 3Medical University of Vienna, Department of Plastic,
Reconstructive, and Aesthetic Surgery, Währinger Gürtel 18-20, 1090 Vienna, Austria; 4University of Hamburg, Department of Clinical
Pharmacology, Institute of Pharmacy, Bundesstrasse 45, 20146 Hamburg, Germany; 5University of Regensburg, Institute of Pharmacy,
Universitätsstraße 31, 93053 Regensburg, Germany; 6Medical University of Vienna, Department of Anesthesiology and General Intensive
Care, Währinger Gürtel 18-20, 1090 Vienna, Austria

*Corresponding author. E-mail: Markus.zeitlinger@meduniwien.ac.at

Received 20 July 2023; accepted 18 November 2023

Background: Pathophysiological changes in severely burned patients alter the pharmacokinetics (PK) of anti-in­
fective agents, potentially leading to subtherapeutic concentrations at the target site. Albumin supplementa­
tion, to support fluid resuscitation, may affect pharmacokinetic properties by binding drugs. This study aimed
to investigate the PK of piperacillin/tazobactam in burn patients admitted to the ICU before and after albumin
substitution as total and unbound concentrations in plasma.
Patients and methods: Patients admitted to the ICU and scheduled for 4.5 g piperacillin/tazobactam adminis­
tration and 200 mL of 20% albumin substitution as part of clinical routine were included. Patients underwent IV
microdialysis, and simultaneous arterial plasma sampling, at baseline and multiple timepoints after drug
administration. PK analysis of total and unbound drug concentrations under steady-state conditions was per­
formed before and after albumin supplementation.
Results: A total of seven patients with second- to third-degree burns involving 20%–60% of the total body
surface were enrolled. Mean (SD) AUC0–8 (h·mg/L) of total piperacillin/tazobactam before and after albumin sub­
stitution were 402.1 (242)/53.2 (27) and 521.8 (363)/59.7 (32), respectively. Unbound mean AUC0–8 before and
after albumin supplementation were 398.9 (204)/54.5 (25) and 456.4 (439)/64.5 (82), respectively.
Conclusions: Albumin supplementation had little impact on the PK of piperacillin/tazobactam. After albumin
supplementation, there was a numerical increase in mean AUC0–8 of total and unbound piperacillin/tazobac­
tam, whereas similar Cmax values were observed. Future studies may investigate the effect of albumin supple­
mentation on drugs with a higher plasma protein binding.

Introduction with a very high mortality.4,5 Hence, beside other intensive care
procedures, infection control has been found to be crucial for
The therapeutic management of patients with severe burns is patient recovery and survival. Inadequate treatment of systemic
one of the most challenging medical conditions in an ICU, requir­
infections has been shown to result in multiple organ dysfunction
ing a multidisciplinary and interdisciplinary team of highly skilled
syndrome and subsequent death.6–9 Studies have shown that
specialists within the field of surgery, intensive care, infectiology
and nutrition.1,2 Severe medical conditions often lead to life- most bacterial infections are caused by Staphylococcus aureus,
threatening infections. Due to a loss of skin barrier and an im­ Pseudomonas aeruginosa and Acinetobacter spp., whereas
munocompromised state, burn patients are at constant risk for Candida spp., Aspergillus spp. and Fusarium spp. account for
bloodstream infections, such as sepsis,3 which are associated most of the fungal infections.4,5

© The Author(s) 2023. Published by Oxford University Press on behalf of British Society for Antimicrobial Chemotherapy.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://
creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the
original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com
262
Protein binding of pip/taz in severely burned patients

Due to unique pathophysiological changes as well as thera­ Study population and procedures
peutic interventions in critically ill patients (e.g. haemodynamic Key inclusion criteria included male and female patients, clinical indica­
changes, organ failure, capillary leakage, extracorporeal therap­ tion for antibiotic treatment with piperacillin/tazobactam and albumin
ies), particularly in those with major burns, the pharmacokinetics substitution as part of clinical routine, ≥18 years of age and admission
(PK) of many anti-infective agents might be affected. This may to the BICU due to major burn injury. All participants were sedated and
lead to changes in drug clearance, volume of distribution (Vd) under mechanical ventilation due to respiratory insufficiency.
and protein binding, conditions potentially resulting in a subopti­ Exclusion criteria were defined as follows: liver and/or kidney failure
mal drug exposure.10–13 Severe burn injuries are associated with requiring organ support, such as renal replacement therapy (RRT) and/
a reduction in plasma albumin levels, which may affect the PK of or extracorporeal membrane oxygenation (ECMO) and pregnancy (apply­
ing to female subjects). For safety reasons, subjects with seropositivity for
protein-bound drugs, as only the unbound fraction of an antibiot­
HIV, HBV (HepB antigen) or HCV (HepC antibody) were not included.
ic exerts a pharmacological effect.14–16 However, the magnitude
Each included individual was scheduled to undergo two consecutive
of impact of change in albumin levels on the dynamically chan­ study days. On Study Day 1, subjects received antibiotic drug treatment
ging concentration-versus-time profile of unbound drugs is still with piperacillin/tazobactam prior to plasma sampling. On Study Day 2,
unknown. albumin substitution of 200 mL of 20% human albumin was infused
Several techniques have been introduced to investigate pro­ over 30 min and followed by the next therapeutic antibiotic drug admin­
tein binding and its effect on antibiotic susceptibility in vivo and istration. Time between administration of albumin and piperacillin/tazo­
in vitro.15,17 In the past years, microdialysis (MD) has been used bactam was consistent throughout all patients.
as a minimally invasive and valuable research tool in the field
of PK analysis, which allows for continuous measurement of un­ Plasma sampling
bound (free) drug concentrations in plasma, as well as in tissues, Piperacillin 4 g/tazobactam 0.5 g was administered IV via a central ven­
to assess protein binding in vivo.18–20 ous catheter over 30 min using a volumetric infusion pump. Blood sam­
The aim of the present study was to detect PK alterations ples were obtained via an arterial catheter. All catheters were
of piperacillin/tazobactam in severely burned patients admit­ implanted by the responsible intensive care clinician as part of clinical
ted to the ICU. Further, we investigated protein binding by routine. For the measurement of the total drug concentrations in plasma,
means of in vivo MD before and after additional albumin blood samples were collected at baseline (prior to next antibiotic admin­
supplementation. istration), at 0.5 (end of infusion), 1, 2, 4, 6 and 8 h after drug infusion.
Plasma sampling on Study Day 2 was performed as on Study Day 1.

Materials and methods MD

Ethics MD is a minimally invasive method for the measurement of free drug con­
centrations in blood and tissues, described in detail elsewhere.18,21–23 MD
This prospective, open-labelled, single-centre study was approved by the
probes with a membrane length of 30 mm and a molecular weight cut-
responsible Ethics Committee of the Medical University of Vienna before
off of 20 kDa were inserted IV and connected to a microinfusion precision
initiation (EC number: 1608/2017) and the Austrian Agency for Health
pump. The system was then continuously perfused with a 0.9% saline
and Food Safety. Further, the study was registered under the EudraCT
solution at a flow rate of 2.0 µL/min. Due to the low molecular weight
number 2017-002216-14.
cut-off of the membrane, only the unbound fraction of drugs was able
All related study procedures were performed at the Burn ICU (BICU) at
to diffuse through and was ultimately collected in the dialysate.18,24
the Medical University of Vienna, Austria, in accordance with the
Samples were calibrated by means of retrodialysis. Two retrodialysis
International Conference on Harmonisation—Good Clinical Practice
samples were collected after MD sampling for each patient and each
(ICH-GCP) guidelines and the Declaration of Helsinki.
study day. The probe was perfused with a solution containing piperacil­
The study population included intubated and unconscious intensive
lin/tazobactam (300/37.5 µg/mL) at a flow rate of 2.0 µL/min, to collect
care patients. Therefore, all procedures and sampling were performed be­
two retrodialysis samples over a period of 30 min each.
fore informed consent was available. However, consent was obtained
Assuming that the diffusion of drugs past the membrane is equal in
from each patient once the subjects were awake and able to understand
both directions, the fraction of the drug concentration in the dialysate
the full content of the study. The actual sample analysis was carried out
(relative recovery) is calculated using the following equations:
after consent was given.

Recovery(%)=100 − (100 × analyteconcentrationout

Study medication and equipment /analyteconcentrationin ).
Included patients received piperacillin/tazobactam 4.5 g (Fresenius Kabi
Austria GmbH, Graz, Austria) in 100 mL of 0.9% saline solution Accordingly, the antibiotic drug concentrations were calculated as follows:
(Fresenius Kabi Austria GmbH) over 30 min every 8 h) as part of routine
care. To ensure steady-state conditions, at least three individual doses Individual drug concentration=100 × (sampleconcentration
were administered prior to the first study day. All study participants
/relativerecovery).
underwent regular kidney function testing and showed mean (SD) glom­
erular filtration rates (GFRs) above 40 mL/min [110.76 (43.12); data not
shown]. Hence, no antibiotic dose adjustment was necessary.
Further, the same substances were used for MD calibration and perfu­ MD sampling
sion. Human albumin 20% for supplementation was purchased from CLS Simultaneous to blood sampling, intravenous MD was used to determine
Behring GmbH, Vienna, Austria. IV MD catheters and microinfusion pumps unbound drug concentrations in plasma. After an equilibration period of at
(107 microdialysis pump) were supplied by M Dialysis (Stockholm, least 30 min, MD sampling was initiated at baseline (prior to next antibiotic
Sweden). drug administration) and at 0–0.5, 0.5–1, 1–2, 2–3, 3–4, 4–5, 5–6, 6–7 and

263
Wulkersdorfer et al.

7–8 h after drug administration. After termination of active sampling, all through PsN 5.0.0.26 One- and two-compartment models with linear dis­
MD probes were calibrated by retrodialysis. Upon completion of all study position and elimination were evaluated to describe the plasma PK of
procedures, the MD probes and venous cannula were removed. All other piperacillin. To model the unbound concentrations from in vitro ultrafiltra­
catheters necessary for further clinical monitoring and/or intensive care tion and in vivo MD, a constant unbound fraction of the total plasma con­
treatment were kept in place until removed by the responsible clinicians. centration was assumed. MD data were modelled using an approach
similar to Tunblad et al.27 to account for interval-based sampling.
Sample handling Inter-individual variability (IIV) of the pharmacokinetic model para­
meters was assumed to be log-normally distributed. For the residual un­
Blood samples were collected in Vacuette® lithium-heparin tubes explained variability in plasma, ultrafiltrate and microdialysate samples,
(Greiner Bio-one, Austria), centrifuged at +4°C and 2600 g for 10 min an additive, proportional or combined error model was evaluated.
and aliquoted in polypropylene Nunc Cryotube™ vials (Thermo Candidate models were evaluated by graphical and numerical criteria
Scientific, Denmark) within 60 min from collection. MD samples were col­ [goodness-of-fit plots, visual predictive checks and drop of objective func­
lected in microvials (M Dialysis) and stored at approximately −20°C. At the tion value (dOFV)]. Parameter uncertainty was evaluated using a
end of each study day all samples were transferred to a −80°C freezer un­ log-likelihood profiling-based sampling-importance resampling routine
til further analysis. (LLP-SIR), a technique for evaluating parameter uncertainty in small
datasets.28
Sample analysis—HPLC-UV method and ultrafiltration The best fitting model for piperacillin was utilized to perform Monte
Carlo simulations to evaluate the suitability of the employed dosing regi­
The concentrations of piperacillin and tazobactam were determined by
men to attain a PK/PD target of 50% and 100% time of the dosing interval
HPLC-UV using a Prominence LC20 modular HPLC system equipped with
that unbound drug concentrations exceed the MIC ( fT>MIC)
an SPD-M30A PDA detector (set to 225 nm for piperacillin or 220 nm for
MICs cannot be determined for tazobactam because it does not exert
tazobactam) and LabSolutions software (Shimadzu, Duisburg, Germany).
antibacterial activity on its own. Nevertheless, a threshold drug concen­
Piperacillin was determined using a NUCLEOSHELL RP18 2.7 µm 100 ×
tration is needed to effectively neutralize the β-lactamase enzymes.
3 mm column for plasma or a CORTECS T3 2.7 µm 100 × 3 mm column for
The simulated unbound concentrations were used for calculation of a
microdialysate (Waters, Eschborn, Germany) preceded by a guard column
PK/PD target of 50% or 100% time of the dosing interval that exceeds
(NUCLEOSHELL RP18 2.7 µm 4 × 3 mm, Macherey-Nagel, Düren, Germany).
the threshold concentration (Ct) ( fT>Ct).
The mobile phase was 0.02 M NaH2PO4/acetonitrile 80:20 (v/v), pH 7.0–
7.1. At a flow rate of 0.4 mL/min and a column temperature of 40°C, pipera­
cillin eluted after 3.2 (NUCLEOSHELL) or 4.8 (CORTECS) min. Sample prepar­
ation for analysis of piperacillin and tazobactam, as well as determination of Results
tazobactam, was performed as previously described.25 In brief, total drug
concentrations were determined after protein precipitation with acetonitrile Study population and study outcome
and removal of acetonitrile by extraction into dichloromethane. Free drug
During a study period from June 2018 to December 2019 a total
concentrations were determined after ultrafiltration using plasma
(300 µL) buffered with 3 M potassium phosphate, pH 7.4 (10 µL), and of seven patients with severe burn injuries admitted to the BICU
Vivafree™ 500, 30 kD Hydrosart® centrifugal ultrafiltration device were enrolled in the study. All participants experienced second-
(Vivaproducts Inc., Littleton, MA, USA). Microdialysate was injected directly. to third-degree burns affecting 20%–60% of the total body sur­
The linearity in plasma was shown from 0.1 to 300 mg/L (R > 0.9982) for face area (TBSA). Albumin supplementation was indicated in six
piperacillin and from 0.125–37.5 mg/L (R > 0.9993) for tazobactam. The re­ out of seven patients, due to low albumin values. However, in
spective values in saline as surrogate for microdialysate or ultrafiltrate were two patients, a defective MD catheter was identified after the se­
R > 0.9996 (concentration 0.03–300 mg/L) for piperacillin and R > 0.9991 cond study day. Hence, due to a lack of reliability, these datasets
(concentration 0.0375–37.5 mg/L) for tazobactam. Based on in-process were removed from further analysis. The final analysis included
quality controls (QCs), the intra- and inter-assay imprecision of the deter­
seven patients on Study Day 1 and four subjects on Study Day 2.
mination of total piperacillin/tazobactam in plasma was <2%/<4% (coeffi­
Table 1 and Table S1 (available as Supplementary data at JAC
cient of variation, CV), the relative error in accuracy was <4%/<1%. The
mean (SD) unbound fraction (fu = Cfree/Ctotal × 100%) of piperacillin/tazobac­ Online) describe the characteristics of the sample. All patients
tam in these QCs was 86.7% (1.9%)/89.2% (5.3%), corresponding to an were sedated and mechanically ventilated. No signs of sepsis
inter-assay precision of 2.2%/6.0% (CV). Based on in-process QCs in saline, were detected by the ICU medical staff.
the intra- and inter-assay precision was <4%/<8% (CV); the accuracy was Overall, patient characteristics did not reveal essential differ­
100.7%/101.0%. ences between Study Days 1 and 2. Mean (SD) albumin levels in­
creased from 24.9 (3.6) g/L on Study Day 1 to 29.5 (2.3) g/L after
PK analysis albumin infusions on Study Day 2. Patients received a mean (SD)
PK data were analysed with a commercially available computer software of 15 (10) single infusions of 4.5 g piperacillin/tazobactam, thus
program (Phoenix® WinNonlin® Build 8.0, Certara USA, Inc., Princeton, NJ, reaching steady-state conditions prior to first study day.
USA) applying a non-compartmental analysis. Where applicable, data Figure 1 illustrates the mean linear steady-state plasma con­
were expressed as mean values. centration–time profiles for bound and unbound piperacillin/
Maximum plasma concentration (Cmax), time to maximum plasma tazobactam, respectively.
concentration (Tmax), elimination half-life (t½), Vd, CL and AUC from 0 to PK parameters of total and free piperacillin and tazobactam
8 h (AUC0–8) were calculated for piperacillin and tazobactam in plasma were expressed as mean (SD) and listed in Tables 2 and 3,
and intravascular dialysate fluid (recovery corrected). respectively. Comparable mean (SD) Cmax values were observed
before and after albumin substitution for total piperacillin
Pharmacometric data analysis [229.11 (53.04) versus 221.26 (65.44) mg/L)] and tazobactam
All quantitative data were analysed using non-linear mixed effects mod­ [25.84 (6.42) versus 23.33 (4.71) mg/L]. Similar results were ob­
elling in NONMEM (ICON, Gaithersburg, MD, USA, version 7.5.0) controlled served for peak free piperacillin [158.1 (49.22) versus 153.6

264
Protein binding of pip/taz in severely burned patients

Table 1. Patient characteristics

Overall (n = 7) Study Day 1 (before albumin) (n = 7) Study Day 2 (after albumin) (n = 6)

Demographic characteristics
Age (years), mean (SD) 52.4 (19.9)
Female sex, n (%) 2 (28.6)
BMI, mean (SD) 29.6 (3.8)
Burn characteristics
Days after burn injury, mean (SD) 6.5 (4.4) 8 (3.0)
TBSA (%), n (%)
20–30 4 (57.1)
30–40 1 (14.3)
40–50 1 (14.3)
50–60 1 (14.3)
TBSA (%), mean (SD) 32.4 (11.4)
Burn degree: II–III, n (%) 7 (100)
No. of piperacillin/tazobactam applications, mean (SD) 15 (10) 20 (9.9)
Laboratory
Albumin (g/L), mean (SD) 24.9 (3.6) 29.5 (2.3)
Mechanical ventilation 7 (100) 6 (100)
Endotracheal tube, n (%) 4 (57.1) 3 (50)
Tracheostomy, n (%) 3 (42.9) 3 (50)
Patient positioning
Sand bed, no (%) 2 (28.6) 1 (16.7)
Alternating pressure system, n (%) 5 (71.4) 5 (83.3)

(102.48) mg/L] and tazobactam [19.47 (7.06) versus 18.03 Figure 2 depicts the resulting PTA plots. For piperacillin, at
(12.21) mg/L]. 100% fT>MIC, high PTA (>90%) was achieved at an MIC value of
Mean (SD) AUC0–8 was numerically higher after albumin sub­ 0.5 mg/L before and 1 mg/L after albumin supplementation.
stitution for total piperacillin [402.08 (242.35) versus 521.8 For 50% fT>MIC, high PTA was obtained for MIC values of 8 mg/L.
(363.01) mg/L) and tazobactam [53.16 (26.78) versus 59.7 For tazobactam, at 100% fT>Ct, high PTA (>90%) was attained
(31.63) mg/L]. Similarly, the AUC0–8 of unbound piperacillin at a Ct of 0.125 before and 0.25 mg/L after albumin supplemen­
[398.92 (203.9) versus 456.35 (438.76) mg/L] and tazobactam tation. For 50% fT>Ct, high PTA was observed at a Ct of 1 mg/L.
[54.48 (25.34) versus 64.49 (82.18) mg/L) increased after albu­
min supplementation.
One implausible outlier led to a strong distortion of t½, Vd and Discussion
CL of free piperacillin/tazobactam before albumin substitution.
Hence, since these PK parameters are highly susceptible to distor­ The present study is the first to investigate PK of anti-infective
tion based on single outliers, we excluded this patient from the agents in severely burned patients before and after albumin sup­
analysis for these variables only. The excluded patient did not plementation, and further to determine the unbound fraction of
undergo Study Day 2. piperacillin/tazobactam in plasma by means of MD. Our objective
was to examine whether an increase in albumin concentration
resulting from therapeutic supplementation could lead to a
Pharmacometric data analysis greater bound fraction of the drug and influence antibiotic drug
A detailed description of the pharmacometric data analysis can exposure.
be found in the Supplementary data. A total of 209 samples Our results show slightly higher mean antibiotic bound and
(plasma: 77, ultrafiltrate: 22, microdialysate: 110) of each drug unbound plasma AUC0–8 values after albumin substitution.
were available for analysis. Within a patient, parameters for CL, Comparable plasma Cmax values for both bound and unbound
fraction unbound from ultrafiltration and recovery-corrected piperacillin/tazobactam were observed. Overall, mentioned
MD data were modelled separately before and after the supple­ data did not show statistical significance (data not illustrated).
mentation with albumin: The best-fitting models for piperacillin As a possible explanation we consider the fact that compared
and tazobactam were utilized to predict unbound plasma con­ with other antibiotic agents, the plasma protein binding (PPB)
centrations through Monte Carlo simulations (n = 500) for the of piperacillin/tazobactam (approximately 30%) is known to be
dosing regimen of 4 g piperacillin and 0.5 g tazobactam every fairly low.29 Hence, antibiotics showing a higher PPB, such as clin­
8 h (Figures S1–S4). Pharmacokinetic and protein-binding model damycin and ceftriaxone, all of which are also selectively used in
parameter estimates for piperacillin and tazobactam can be the BICU, may be affected differently.8 However, our choice of
found in Tables S2 and S3, respectively. antibiotic was based on prescription frequencies in ICU units, to

265
Wulkersdorfer et al.

Figure 1. Mean (SD) concentration–time curves of total (a) and unbound (b) piperacillin and total (c) and unbound (d) tazobactam concentrations in
plasma of burn patients after an IV application of 4.5 g of piperacillin/tazobactam at steady-state conditions before and after albumin substitution.
This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

provide insights into PPB characteristics of more commonly used Accordingly, there was a slight decrease in clearance for both
drugs. bound and unbound piperacillin and total tazobactam, but not
Piperacillin is subject to filtration and substantial tubular se­ for free tazobactam after albumin substitution. Nevertheless,
cretion in the renal tubules. The unbound fraction is mainly fil­ these results should be interpreted with caution, as the increase
tered, while the bound fraction can be eliminated through in AUC was not significant and the variability was high.
tubular secretion.30 The simultaneous increase in both unbound Severe burn trauma is associated with unique pathophysio­
and bound fractions of piperacillin/tazobactam therefore indi­ logical changes, such as a hypermetabolic response, which initi­
cates a simultaneous and proportional rise in filtration and secre­ ates >48 h after initial burn trauma, and is proportional to the
tion. Despite the consistency in Cmax before and after albumin size of the burn.31 The mean (SD) number of days from burn injury
supplementation, the increase in AUC for both bound and un­ to study initiation was 6.5 (4.4) and the affected %TBSA was
bound fractions implies a shift in the drug’s clearance. >20% in all patients. As a result, all subjects were expected to

266
Protein binding of pip/taz in severely burned patients

Table 2. Key pharmacokinetic parameters of piperacillin in plasma (total concentrations) and microdialysate (free concentrations) after an IV infusion
of 4.5 g piperacillin/tazobactam at steady-state conditions before and after albumin substitution

Plasma before albumin Plasma after albumin MD before albumin MD after albumin
Piperacillin (4 g) (total concentration) (total concentration) (free concentration) (free concentration)

Cmax (mg/L) 229.11 (53.04) 221.26 (65.44) 158.10 (42.22) 153.6 (102.48)
AUClast (hamg/L) 402.08 (242.35) 521.80 (363.01) 398.94 (203.92) 492.46 (510.815)
Tmax (h) 0.5 0.5 0.68 (0.19) 0.75
t½ (h) 2.58 (1.87) 2.77 (1.85) 2.01 (1.8)a 2.95 (2.16)
CL (L/h) 11.75 (6.16) 10.14 (6.76) 12.55 (6.58)a 12.44 (7.72)
Vd (L) 31.71 (9.16) 27.09 (5.73) 26.37 (10.11)a 35.22 (14.03)

Values are expressed as mean (SD).

a
Results after exclusion of one outlier (n = 6).

Table 3. Key pharmacokinetic parameters of tazobactam in plasma (total concentrations) and microdialysate (unbound concentrations) after an IV
infusion of 4.5 g piperacillin/tazobactam at steady-state conditions before and after albumin substitution

Plasma before albumin Plasma after albumin MDs before albumin MD after albumin
Tazobactam (0.5 g) (total concentration) (total concentration) (free concentration) (free concentration)

Cmax (mg/L) 25.84 (6.42) 23.33 (4.72) 19.47 (7.06) 18.04 (12.21)
AUClast (hamg/L) 53.16 (26.78) 59.70 (31.63) 54.48 (25.34) 64.49 (64.40)
Tmax (h) 0.5 0.5 0.75 0.75
t½ (h) 2.70 (2.00) 2.93 (1.92) 2.34 (2.14)a 3.21 (2.45)
CL (L/h) 10.22 (5.35) 9.19 (4.67) 10.38 (4.76)a 11.09 (6.31)
Vd (L) 28.89 (9.43) 28.20 (3.97) 25.12 (8.63)a 34.89 (15.31)

Values are expressed as mean (SD).

a
Results after exclusion of one outlier (n = 6).

be in the hypermetabolic phase during the study period. In this and 8 mg/L, respectively. These breakpoints serve as a reference
state, the metabolism of anti-infective agents may be altered for susceptibility and can guide empirical antibiotic coverage.
due to factors such as increased cardiac output and hypoalbumi­ In our simulation, after albumin supplementation, the PTA of
naemia. Accordingly, and in agreement with previous studies in­ piperacillin achieving 50% fT>MIC amounted to 81% (16 mg/L)
vestigating PK parameters of piperacillin in severely burned and 92% (8 mg/L) for P. aeruginosa and Enterobacterales,
patients, our subjects showed an increased Vd and prolonged respectively.
t½ when compared with healthy volunteers.29,32 Although in vivo animal studies have demonstrated that
Piperacillin exhibits time-dependent killing, which is achieved β-lactams require target times of 40%–70% fT>MIC, retrospective
when the free fraction of the drug in plasma continuously ex­ analyses suggest that a much larger exposure of 100% fT>4×MIC
ceeds the MIC of the target pathogen. Therefore, clinical efficacy may be necessary to achieve clinical effectiveness.33 Achieving
is best estimated by the percentage of the dosing interval, in these targets with a short infusion may prove to be unattainable
which the free fraction exceeds the MIC of target pathogens and could lead to toxic drug exposures. Therefore, when aiming
(fT>MIC). Assuming a piperacillin dose of 4 g q8h infused over to combat Gram-negative pathogens with elevated MICs, consid­
0.5 h, Monte Carlo simulations predicted a PTA of 50% fT>MIC ering extended or continuous drug infusions may emerge as a vi­
for MIC values of ≤8 mg/L. No significant difference was observed able strategy to optimize drug exposure.
in PTA before or after albumin supplementation. In contrast, We acknowledge the limitation of a relatively small sample
tazobactam does not exert antibacterial properties on its own, size, in which not all subjects underwent complete MD sampling
but restores the activity of piperacillin. Therefore, MICs cannot to assess the unbound fraction. This led to relatively high variabil­
be determined for tazobactam alone, but a Ct must be achieved ity in our results. Furthermore, the pharmacometric data analysis
to effectively neutralize β-lactamase enzymes. Monte Carlo si­ did not evaluate covariate effects beyond albumin, as the study
mulations predicted a PTA of >90% for achieving 50% fT>Ct for size was too small to obtain precise and accurate estimates of
a threshold concentration of 1 mg/L (before and after albumin covariate effects at sufficiently high power. Nevertheless, this
supplementation). was the first study to successfully establish a complete simultan­
Common pathogens of infections in burn intensive care include eous plasma concentration–time profile of total and unbound
P. aeruginosa and Enterobacterales.3–5 According to EUCAST, the fractions of piperacillin and tazobactam and to evaluate the ef­
clinical breakpoints for P. aeruginosa and Enterobacterales are 16 fects of albumin supplementation on their PK. Finally, we

267
Wulkersdorfer et al.

Figure 2. PTA plots for 0.5 h infusion of 4.5 g piperacillin/tazobactam q8h under steady-state conditions. (a) PTA of piperacillin for 50% fT>MIC, (b) PTA of
tazobactam for 50% fT>Ct, (c) PTA of piperacillin for 100% fT>MIC and (d) PTA of tazobactam for 100% fT>Ct. Comparison between concentrations before
albumin supplementation (grey lines) and after albumin supplementation (red lines) with simulated (n = 500) piperacillin/tazobactam concentrations.
Horizontal lines denote breakpoints of 90% PTA. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

acknowledge the limitation of having excluded one patient for Conclusions

the calculation of t½, CL and Vd of free piperacillin/tazobactam
Albumin supplementation had little impact on the PK of piperacil­
before albumin substitution. Nevertheless, given the small sam­ lin/tazobactam. After albumin supplementation, there was a nu­
ple size, we chose to include the data for calculating Cmax, AUC merical increase in mean AUC0–8 of total and unbound
and Tmax of free piperacillin/tazobactam, as well as for generat­ piperacillin/tazobactam, whereas similar Cmax values were ob­
ing the corresponding figures and models, since these para­ served. Future studies may investigate the effect of albumin sup­
meters are less influenced by individual outliers. plementation on drugs with a higher plasma protein binding.

268
Protein binding of pip/taz in severely burned patients

Monte Carlo simulations indicate that dosing regimens of 4 g 2 Lorente JA, Amaya-Villar R. Update in the management of critically ill
piperacillin/0.5 g tazobactam q8h was sufficient for empirical burned patients. Med Intensiva 2016; 40: 46–8. https://doi.org/10.1016/
treatment of common Gram-negative pathogens. j.medin.2015.10.009
3 Brusselaers N, Monstrey S, Snoeij T et al. Morbidity and mortality of
bloodstream infections in patients with severe burn injury. Am J Crit
Care 2010; 19: e81–7. https://doi.org/10.4037/ajcc2010341
Acknowledgements 4 Branski LK, Al-Mousawi A, Rivero H et al. Emerging infections in burns.
We thank the entire BICU of the Medical University of Vienna, especially Surg Infect 2009; 10: 389–97. https://doi.org/10.1089/sur.2009.024
the nursing staff led by Elisabeth Bulant and Diana Raidl, for their active
support and collaboration throughout the study process. 5 Church D, Elsayed S, Reid O et al. Burn wound infections. Clin Microbiol
Rev 2006; 19: 403–34. https://doi.org/10.1128/CMR.19.2.403-434.2006
6 Greenhalgh DG. Management of burns. NEJM 2019; 380: 2349–59.
https://doi.org/10.1056/NEJMra1807442
Funding 7 Greenhalgh DG, Saffle JR, Holmes JH et al. American Burn Association
This project was internally funded by the Department of Clinical Consensus Conference to define sepsis and infection in burns. J Burn Care
Pharmacology at the Medical University of Vienna. All mentioned supplies Res 2007; 28: 776–90. https://doi.org/10.1097/BCR.0b013e3181599bc9
were provided by the Department of Clinical Pharmacology and in part by
8 Wang Y, Beekman J, Hew J et al. Burn injury: challenges and advances
the Burn Intensive Care Unit (BICU) at the Medical University of Vienna.
in burn wound healing, infection, pain and scarring. Adv Drug Deliv Rev
2018; 123: 3–17. https://doi.org/10.1016/j.addr.2017.09.018
9 Dries DJ, Marini JJ. Management of critical burn injuries: recent devel­
Transparency declarations opments. Acute Crit Care 2017; 32: 9–21. https://doi.org/10.4266/kjccm.
None to declare. 2016.00969
10 Roberts JA. Using PK/PD to optimize antibiotic dosing for critically ill
patients. Curr Pharm Biotechnol 2011; 12: 2070–9. https://doi.org/10.
Author contributions
2174/138920111798808329
All authors have declared full responsibility for the work described in this 11 Roberts JA, Lipman J. Pharmacokinetic issues for antibiotics in the
manuscript and have provided final approval of the version submitted to critically ill patient. Crit Care Med 2009; 37: 840–51. https://doi.org/10.
the journal. Beatrix Wulkersdorfer planned and performed the clinical 1097/CCM.0b013e3181961bff
study, interpreted the data, created tables and figures and wrote the
manuscript. Felix Bergmann interpreted the data, created tables and fig­ 12 Jaehde U, Sorgel F. Clinical pharmacokinetics in patients with burns.
ures and wrote the manuscript. Lisa Amann performed pharmacokinetic Clin Pharmacokinet 1995; 29: 15–28. https://doi.org/10.2165/00003088-
modelling, created tables and figures and critically revised the manu­ 199529010-00003
script. Alexandra Fochtmann-Frana performed the clinical study and crit­ 13 Jager NG, van Hest RM, Lipman J et al. Therapeutic drug monitoring of
ically revised the manuscript. Valentin al Jalali performed the clinical anti-infective agents in critically ill patients. Expert Rev Clin Pharmacol
study and critically revised the manuscript. Elizaveta Kurdina performed 2016; 9: 961–79. https://doi.org/10.1586/17512433.2016.1172209
the clinical study and critical revised the manuscript. Edith Lackner 14 Jamal JA, Economou CJ, Lipman J et al. Improving antibiotic dosing in
planned and performed the clinical study. Sebastian G. Wicha performed special situations in the ICU: burns, renal replacement therapy and extra­
pharmacokinetic modelling, created tables and figures and critically re­ corporeal membrane oxygenation. Curr Opin Crit Care 2012; 18: 460–71.
vised the manuscript. Christoph Dorn performed the bioanalysis and crit­ https://doi.org/10.1097/MCC.0b013e32835685ad
ically revised the manuscript. Bruno Schäfer planned the clinical study
15 Beer J, Wagner CC, Zeitlinger M. Protein binding of antimicrobials:
and critically revised the manuscript. Gerald Ihra planned the clinical
methods for quantification and for investigation of its impact on bacterial
study and critically revised the manuscript. Thomas Rath planned the
killing. AAPS J 2009; 11: 1. https://doi.org/10.1208/s12248-008-9072-1
clinical study and critically revised the manuscript. Christine Radtke
planned the clinical study and critically revised the manuscript. Markus 16 Acharya C, Hookers AC, Turkyilmaz GY et al. A diagnostic tool for popu­
Zeitlinger planned and performed the clinical study, interpreted the lation models using non-compartmental analysis: the ncappc package
data and critically revised the manuscript. for R. Comput Methods Programs Biomed 2016; 127: 83–93. https://doi.
org/10.1016/j.cmpb.2016.01.013
17 Sebille B. Methods of drug protein binding determinations. Fundam
Clin Pharmacol 1990; 4 Suppl 2: 151s–61s. https://doi.org/10.1111/j.
Data and materials availability 1472-8206.1990.tb00073.x
The data and materials that support the findings of this study are avail­ 18 Muller M, Schmid R, Georgopoulos A et al. Application of microdialysis
able from the corresponding author upon reasonable request. to clinical pharmacokinetics in humans. Clin Pharmacol Ther 1995; 57:
371–80. https://doi.org/10.1016/0009-9236(95)90205-8
19 Elmquist WF, Sawchuk RJ. Application of microdialysis in pharmacoki­
Supplementary data netic studies. Pharm Res 1997; 14: 267–88. https://doi.org/10.1023/A:
Pharmacometric data analysis, Figures S1 to S4 and Tables S1 to S3 are 1012081501464
available as Supplementary data at JAC Online. 20 Brunner M, Langer O. Microdialysis versus other techniques for the
clinical assessment of in vivo tissue drug distribution. AAPS J 2006; 8:
E263–71. https://doi.org/10.1007/BF02854896
21 Burian A, Wagner C, Stanek J et al. Plasma protein binding may reduce
References antimicrobial activity by preventing intra-bacterial uptake of antibiotics,
1 Rex S. Burn injuries. Curr Opin Crit Care 2012; 18: 671–6. https://doi.org/ for example clindamycin. J Antimicrob Chemother 2011; 66: 134–7.
10.1097/MCC.0b013e328359fd6e https://doi.org/10.1093/jac/dkq400

269
Wulkersdorfer et al.

22 Zeitlinger MA, Derendorf H, Mouton JW et al. Protein binding: do we 28 Broeker A, Wicha SG. Assessing parameter uncertainty in small-n
ever learn? Antimicrob Agents Chemother 2011; 55: 3067–74. https:// pharmacometric analyses: value of the log-likelihood profiling-based
doi.org/10.1128/AAC.01433-10 sampling importance resampling (LLP-SIR) technique. J Pharmacokinet.
23 Verbeeck RK. Blood microdialysis in pharmacokinetic and drug me­ Phar 2020; 47: 219–28. https://doi.org/10.1007/s10928-020-09682-4
tabolism studies. Adv Drug Deliv Rev 2000; 45: 217–28. https://doi.org/ 29 Sorgel F, Kinzig M. Pharmacokinetic characteristics of piperacillin/
10.1016/S0169-409X(00)00110-1 tazobactam. J. Intensive Care Med 1994; 20 Suppl 3: S14–20. https://
24 Muller M. Science, medicine, and the future: microdialysis. BMJ 2002; doi.org/10.1007/BF01745246
324: 588–91. https://doi.org/10.1136/bmj.324.7337.588 30 Currie GM. Pharmacology, part 2: introduction to pharmacokinetics. J
25 Kratzer A, Schiesser S, Matzneller P et al. Determination of total and Nucl Med Technol 2018; 46: 221–30. https://doi.org/10.2967/jnmt.117.
free ceftolozane and tazobactam in human plasma and interstitial fluid 199638
by HPLC-UV. J Pharm Biomed Anal 2019; 163: 34–8. https://doi.org/10. 31 Jeschke MG, Mlcak RP, Finnerty CC et al. Burn size determines the in­
1016/j.jpba.2018.09.044 flammatory and hypermetabolic response. J Crit Care 2007; 11: R90.
26 Lindbom L, Pihlgren P, Jonsson EN. PsN-Toolkit–a collection of com­ https://doi.org/10.1186/cc6102
puter intensive statistical methods for non-linear mixed effect modeling 32 Jeon S, Han S, Lee J et al. Population pharmacokinetic analysis of
using NONMEM. Comput Methods Programs Biomed 2005; 79: 241–57. piperacillin in burn patients. Antimicrob Agents Chemother 2014; 58:
https://doi.org/10.1016/j.cmpb.2005.04.005 3744–51. https://doi.org/10.1128/AAC.02089-13
27 Tunblad K, Hammarlund-Udenaes M, Jonsson EN. An integrated mod­ 33 Roberts JA, Paul SK, Akova M et al. DALI: defining antibiotic levels in
el for the analysis of pharmacokinetic data from microdialysis experi­ intensive care unit patients: are current β-lactam antibiotic doses suffi­
ments. Pharm Res 2004; 21: 1698–707. https://doi.org/10.1023/B:PHAM. cient for critically ill patients? Clin Infect Dis 2014; 58: 1072–83. https://
0000041468.00587.c6 doi.org/10.1093/cid/ciu027

270