Documente Academic
Documente Profesional
Documente Cultură
A microbiological method for determining serum levels of broad spectrum
-lactam antibiotics in critically ill patients
PII: S0167-7012(16)30192-0
DOI: doi: 10.1016/j.mimet.2016.07.020
Reference: MIMET 4966
Please cite this article as: Fridlund, Jimmy, Woksepp, Hanna, Schon, Thomas,
A microbiological method for determining serum levels of broad spectrum -lactam
antibiotics in critically ill patients, Journal of Microbiological Methods (2016), doi:
10.1016/j.mimet.2016.07.020
This is a PDF le of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its nal form. Please note that during the production process
errors may be discovered which could aect the content, and all legal disclaimers that
apply to the journal pertain.
ACCEPTED MANUSCRIPT
T
Jimmy Fridlund1, Hanna Woksepp1,2, Thomas Schn1,2,3*
IP
1
Department of Clinical Microbiology, Kalmar County Hospital, Kalmar, Sweden
2
R
Department of Medicine and Optometry, Linnaeus University, Kalmar, Sweden.
3
SC
Department of Medical Microbiology, Linkping University, Linkping, Sweden.
NU
MA
D
TE
REVISED VERSION
P
CE
AC
*Corresponding author:
Thomas Schn
Dept. of Clinical Microbiology, Kalmar County Hospital
SE-391 85 Kalmar, Sweden
Tel: +4648081000, Email:tschon@hotmail.com
1
ACCEPTED MANUSCRIPT
Abstract
Background: Recent studies show that suboptimal blood levels of -lactam antibiotics are
present in intensive care unit (ICU) patients. A common reference method for assessing drug
T
highly accurate but rarely available outside reference centres. Thus, our aim was to develop a
IP
microbiological method for monitoring -lactam antibiotic serum levels which could be used
R
SC
at any hospital with a microbiological laboratory.
Methods: The method was developed as a 96-well broth microdilution format to assess the
NU
concentrations of cefotaxime (CTX), meropenem (MER), and piperacillin (PIP). Patient
serum containing antibiotics were diluted in suspensions of bacteria with known minimal
MA
inhibitory concentrations (MICs). Serum antibiotic concentrations were calculated by dividing
the MIC with the dilution factor at which the serum inhibited growth of the bacterial
D
suspension. Serum (n=88) from ICU patients at four hospitals in south-east Sweden were
TE
Results: The overall accuracy and precision for spiked samples and patient samples was
CE
within the pre-set target of 20.0% for all drugs. There was a significant correlation between
AC
the microbiological assay and LC-MS for the patient samples (CTX: r= 0.86, n=31; MER: r=
0.96, n=11; PIP: r= 0.88, n=39) and the agreement around the clinical cut-off for CTX (4.0
mg/l), MER (2.0 mg/l) and PIP (16.0 mg/l) was 90%, 100% and 87%, respectively.
levels of meropenem, piperacillin and cefotaxime suitable for clinical use. It is an inexpensive
2
ACCEPTED MANUSCRIPT
Introduction
The suggested target concentration for -lactam antibiotics in patients who are not critically ill or
immunosuppressed is at least 50% of the exposure time (T) for the free (f) fraction above the
T
minimal inhibitory concentration (MIC) (50% f T>MIC) (Roberts et al., 2014). For patients admitted to
IP
the intensive care unit (ICU), a number of studies indicate that a target of 100% f T1-4x>MIC is required
R
for maximum clinical efficacy and to limit development of resistance (Dulhunty, et al., 2013; Wong
SC
et al., 2014; Donadello et al., 2015). According to the DALI study (Roberts et al., 2014),
NU
concentrations of -lactams below 100% f T>MIC in critically ill patients were present in about 30%
(LC-MS/MS) (Carlier, et al., 2015). Alternative methods, such as immunoassays (Grange et al.,
TE
2014), and microbiological assays are available for some antibiotics, such as vancomycin (Rybak et
P
al., 1991), teicoplanin (Erickson, et al., 1989) and ceftriaxone (Dafale et al., 2012). In general, LC-
CE
MS/MS based methods are used for therapeutic drug monitoring at reference centres (Mller and
Rentsch, 2010; Wong, et al., 2014). This method is highly accurate and able to assess multiple
AC
compounds in a single sample. It is however, not readily available to most hospitals due to high cost
In Europe, susceptibility testing of bacteria by using a MIC determination method has been
harmonized by EUCAST (EUCAST, 2003). A modified version of this method could also be used to
assess antibiotic concentrations from human samples. Some examples are the studies by Steinmann
et al. (2010) and Candejas-Bueno (2012), where the concentration of the antifungal drugs
voriconazole and posaconazole were assessed by adding these drugs to wells in agar plates covered
with microbes. The concentrations were determined by the size of the clear area around the well. To
our knowledge, a microbiological method for assessing the concentrations of -lactam antibiotics for
3
ACCEPTED MANUSCRIPT
quantitative clinical assessment in patient serum samples has not previously been described and
validated. The purpose of a microbiological method would be to achieve a performance adequate for
clinical monitoring of antibiotic concentrations in patients where more accurate LC-MS methods are
T
not available. Due to the increasing need for individualizing antibiotic treatment in critically ill
IP
patients based on drug concentrations, our aim was to develop a microbiological method for
R
monitoring cefotaxime (CTX), meropenem and piperacillin serum levels which could be used at any
SC
hospital with a clinical microbiological laboratory.
NU
Material and Methods
Selection of bacterial strains
MA
The bacterial strains used in the microbiological assay were selected based on the MIC values
quality control ranges established by EUCAST (EUCAST, 2003). The main selection criteria for
D
bacteria were the likelihood of incorporating the MIC when creating dilution series in order to most
TE
accurately identify concentrations around the clinical target MIC of each antibiotic (CTX: 4.0 mg/L;
MER: 2.0 mg/L, and PIP: 16.0 mg/L) (Roberts et al., 2014). Strains were also selected based on their
P
CE
speed of growth, stability and required growth conditions. Two bacterial species were used for CTX
and MER (Table 1); one with a high and one with a low MIC for these drugs in order to cover the
AC
target range for analysis in one run. For PIP, one bacterium was used, and the serum sample was
analysed undiluted and in a 1:2 dilution to cover a wider concentration range. The strains were stored
in aliquots at -70C in cryoinstant tubes (Scharlau; Barcelona, Spain). Before experiments, the
bacteria were grown on blood agar plates for 20 1h under aerobic conditions at 35-37C and
Antibiotic stock solutions were prepared using PIP, MER, and CTX from Sigma Aldrich (St.
Louis, US-MO) (Purity by HPLC: CTX: 100.0%; MER: 99.8%; and PIP: 97.5%) diluted in sterile
distilled water to a final concentration of 5120 mg/l and stored at -70C for a maximum of 12
4
ACCEPTED MANUSCRIPT
months. The internal controls for CTX (4mg/L), MER (2mg/L) and PIP (16mg/L) were diluted in
Mueller-Hinton broth (MHb; DIFCO, Becton Dickinson, USA), aliquoted on a single occasion and
were kept in storage at -70C until used, with a maximum storage period of 12 months.
T
Methodological outline for determining concentrations of cefotaxime, piperacillin and
IP
meropenem by the microbiological assay
R
The microbiological assay was developed in a 96-well, broth-microdilution format. All solutions
SC
and patient samples used were equilibrated to room temperature prior to their use. The selected
NU
bacteria where MICs were predefined for the antibiotic to be tested (Table 1) in which the serum
samples were diluted, were included in each experiment (for experimental plate setup, please refer to
MA
supplementary method: Figure 1). In order to assess the serum antibiotic concentration, serum
samples were diluted using MHb containing the bacteria selected for the antibiotic to be tested
D
(Table 1). After incubation, the concentration of each serum sample was calculated by dividing the
TE
MIC with the serum dilution factor inhibiting bacterial growth. The MIC used for detection was
determined in each experiment using an MIC control dilution series (MICControl) (supplementary
P
CE
Each serum sample was analysed using two separate 96-well plates using two different bacteria to
cover for the high (SerumConc.S.aureus for CTX and SerumConc.P.aeruginosa for MER), and the low
MICs. For PIP, the same bacterium was used for verifying both high and low concentrations
(SerumConc.E.coli-PIP-1:1; SerumConc.E. coli-PIP-1:2) in order to cover the analytical range (4-96 mg/l).
5
ACCEPTED MANUSCRIPT
As controls, both the MIC determination of the bacteria used for the assay (MICControl) as well as
an internal control with a fixed concentration for each antibiotic diluted in MHb were included. To
T
prepare the MICControl, the stock solution of the antibiotic to be determined (5120mg/l) was further
IP
diluted to the starting concentration indicated for the MICControl dilution in supplementary method:
R
Tables 1-5. Then, the MICControl was dispensed in 1.5 ml tubes according to the dilution strategy in
SC
supplementary method: Tables 1-5. For the serum samples, one part was diluted 1:5 in MHb for CTX
NU
and MER and 1:2 for PIP for analysis of the higher concentration ranges. The undiluted part of the
serum sample used for the lower concentration ranges as well as the diluted serum sample and the
MA
undiluted internal control were then dispensed in 1.5 ml tubes (for specific volumes depending on
Before analysis, the bacteria as described in Table 1 were grown on blood agar plates for 201h
P
under aerobic conditions at 35-37C. Bacterial colonies were dissolved in 0.9% NaCl (Merck, NJ,
CE
U.S) to 0.5 McF by using a Densichek plus spectrometer (BioMrieux SA, France). The solution was
subsequently diluted 1:100 in MHb, to provide an approximate final concentration of 1*105 colony
AC
forming units (CFU) (EUCAST, 2003). The bacterial suspensions were then added to the MICControl
and the serum samples according to the dilution schemes in supplementary method: Tables 1-5
Outline for dispensing samples on the 96-well microtiter plate for analysis
Tubes with bacterial suspension and antibiotics including the MICControl (in duplicate) were mixed
vigorously with vortex agitation before 100 l was added to a 96-well microtiter plate (Sarstedt,
without antibiotics was used as positive control. MHb without bacteria was used as negative control.
6
ACCEPTED MANUSCRIPT
Plates were covered with adhesive film (Sigma Aldrich St. Louis, US-MO) and incubated for 201 h
at 35-37C before being analysed by visual inspection to determine the MICs. To pass the quality
control, we predefined that the internal control in each experiments should be within 20.0% of the
T
target (2.8-5.2 mg/l for CTX, 1.4-2.6 mg/l for MER and 11.2-20.8 mg/l for PIP) similar to described
IP
by Candejas-Bueno et al. (2013). Additionally, the MICControl used for each assay had to be within
R
one MIC dilution (EUCAST, 2003). If these conditions were not fulfilled, the analysis results were
SC
discarded and the sample was re-analysed.
NU
The concentration of the patient samples retrieved in each of the two 96-well plates were
calculated by dividing the plate specific MIC (mean of the two MICControl series), with the dilution
MA
factor of the first well lacking growth in the patient sample dilutions. The final concentration of the
serum sample was then determined from the concentrations in each of the two both 96-well plates in
D
accordance with the flow charts presented in supplementary method: Figures 2-4.
TE
In order to estimate the rate of patients achieving 100% f T>MIC, serum samples from ICU patients
County Hospital; Linkping University Hospital; Vxj Central Hospital; and Jnkping County
Hospital Ryhov) were analysed (CTX n=31; MER n=18; PIP n=39). Samples were collected
immediately prior to administration of the next antibiotic dose. The study and the analysis of
antibiotic concentrations in patient serum were approved from the ethical committee at Linkping
University (Dnr 2014/236-31). The final protocol was used to assess the concentration in clinical
samples from ICU patients (CTX: n=31; MER: n=18; and PIP: n=39) stored at -70C for a maximum
of 12 months. The concentrations attained were compared to those attained previously, using LC-MS
in order to evaluate the accuracy and precision of unknown clinical samples. Analysis of PIP, MER
and CTX concentrations by LC-MS was performed at the Department of Clinical Pharmacology,
7
ACCEPTED MANUSCRIPT
accuracy and precision for LC-MS are within +/-15 % for all drugs and the measurement range is
0.2-100 mg/L for PIP and 0.2-50 mg/L for MER and CTX.
T
IP
Statistics
R
Statistical analysis of the correlation between the results of the microbiological assay and the LC-
SC
MS reference assay was done by using Spearman Rank Correlation test, in GraphPad Prism version 6
NU
for Windows, (GraphPad Software, San Diego California USA). In order to assess the accuracy and
precision of the microbiological method, RE% and the CV% was calculated. RE% was calculated
MA
using the sample deviation from the reference-concentration (i.e. LC-MS-concentrations for the
patient serum samples, and the known concentration in the case of spiked samples) divided by the
D
reference-concentration.
P TE
CE
AC
8
ACCEPTED MANUSCRIPT
Results
Stability of MIC determinations for bacterial control isolates throughout the experiments
Throughout the method development and clinical sample analysis, the MICControls were recorded
T
and compared to the MICs pre-defined by EUCAST for the ATCC strains used (Table 1). None of
IP
the antibiotics showed more than a 6% (CTX: n= 12/203; MER: n= 5/221; and PIP: n= 7/235)
R
deviation beyond 1 two-fold MIC-dilution steps from the target MIC range (CTX: 2-8 mg/l; MER:
SC
1-4 mg/l; and PIP: 8-32 mg/l) (Table 1). In case the MIC was out of the target range, the analysis was
NU
declared invalid, and re-analysed.
Performance of the microbiological assay compared to LC-MS for ICU patient serum samples
MA
The overall inter-assay variation RE% and CV% were within the pre-set target of 20% for all
three antibiotics when using spiked serum samples (please refer to validation supplement for details).
D
In clinical serum samples, the comparison between the microbiological assay and LC-MS showed
TE
that the RE% and CV% results for all three antibiotics were within the 20.0% cut-off limit (CTX:
P
CV% = 9.5, RE%: = (-9.7); MER: CV% = 5.5, RE%: = 5.6; and PIP: CV% = 11.1, RE%: = (-19.4),
CE
supplementary validation: Table 1). Comparing the results of the microbiological assay to the
reference method revealed a strong correlation between the two methods for all three antibiotics
AC
(CTX: r= 0.86, p< 0.001, n = 31; MER: r= 0.96, p< 0.05, n = 18; PIP: r= 0.88, p< 0.01, n = 39)
(Figures 1A-C). The agreement between the reference method and the microbiological assay
regarding the concentration assessment around the clinical cut offs were 90 % (CTX = 28/31), 100 %
(MER = 18/18) and 87 % (PIP = 34/39) for the three drugs respectively.
9
ACCEPTED MANUSCRIPT
Discussion
meropenem, and piperacillin which could be performed in any microbiological laboratory at a low
T
cost. The method showed adequate performance within 20.0% for CV% and RE%, both when
IP
considering inter- and intra-assay variation. Thus, the accuracy and precision of the analysis is
R
sufficient for clinical use in recommending dose adjustments if serum drug levels are below the cut-
SC
off.
NU
We found few, if any published microbiological methods for determining serum levels of
cefotaxime or piperacillin for clinical purposes although it was used by one centre in a recently
MA
published multi-centre survey on the use of therapeutic drug monitoring of -lactam antibiotics in
ICU patients (Wong, et al., 2014). For meropenem, a microbiological assay was described (Mendez
D
Compared to the reference microdilution broth MIC method described by the EUCAST in which
P
a 1:2 dilution is used, the microbiological method applies a narrower MIC-dilution, both for the
CE
bacterial isolate for diluting the serum sample as well as for the internal control and serum samples.
Although, the 1:2 dilution is standard for MIC testing in clinical practise, there was a need to create a
AC
more fine-tuned MIC dilution in order to accurately determine serum drug concentrations by using
the known MICs. As the dilution strategy was central in the method development we choose for
simplicity to dilute the serum samples as well as the bacteria. This strategy was regarded as
acceptable as the difference in the bacterial volume did not affect the final MIC-determination as
Even if the microbiological assay has a performance which is inferior to LC-MS for determining
antibiotic concentrations, our results show that it is sufficient for the demands in clinical routine. In
the microbiological assay, we noted that there were systematic trends where CTX and PIP tends to
10
ACCEPTED MANUSCRIPT
Tables 2-3). LC-MS is a more accurate and precise method which should be used where available
and it is also faster than the microbiological method as it can be analysed within the same day. In the
T
microbiological method one drug concentration takes about 1h hands-on time but requires overnight
IP
incubation. However, the microbiological assay could be of importance outside reference centres
R
where LC-MS is not available and sample transport may delay results on critically ill patients for at
SC
least 3-4 days.
NU
There are several reasons why the microbiological method is not as accurate as LC-MS. Small
differences during the preparation of the dilution series providing the MICControl for each plate might
MA
affect the results. Such potential variations might be caused by bacterial clumping and variations in
the bacterial inoculum. This risk is especially likely for S. aureus, as is noted by Cotar et al. (2010),
D
as, due to specialized surface proteins, it is highly prone to clumping. The ability of P. aeruginosa to
TE
form adhesive biofilms may also contribute to higher-than-intended numbers of cells to the MHb-
bacteria mix. Another possibility is that, interference with bacterial growth from compounds in the
P
CE
patient sample, including parts of the immune system present in the sera such as antibodies and
components of the complement system may affect the results. In addition, despite the high
AC
correlation between the two methods (Figures 1A-C), there are occasional differences between the
microbiological assay and the LC-MS method. One explanation for this might be that the samples
had been analysed previously (3-9 months before) by LC-MS and the lower concentrations in the
microbiological assay could be explained by antibiotic degradation. Moreover, the readout in the
microbiological method is based on visual evaluation of the MIC in contrast to LC-MS which is
more objective.
Our method has several limitations. The major one is the inability to assess the drug concentration
of patients being treated using more than one antibiotic. Here, additional treatment with drugs which
affect the susceptibility of the bacteria used in the assay such as fluoroquinolones or aminoglycosides
11
ACCEPTED MANUSCRIPT
may overestimate drug concentrations. Thus, we recommend that the patients should be on single
therapy for more than 24 hours before the microbiological assay is used. This particular limitation
may require some continued work on this method, as multiple antibiotics is recommended as a
T
treatment for severe pneumonia and sepsis (Wong, et al., 2014). In the study by Woksepp et al.
IP
(Submitted), approximately 30% of ICU patients in South-East Sweden were treated with more than
R
one antibiotic.
SC
Nevertheless, the microbiological method could be an important contribution for hospitals without
access to a LC-MS method for determining broad spectrum -lactam drug concentrations in critically
NU
ill patients. However, the method should be prospectively validated against a reference method such
MA
as LC-MS for the purpose of therapeutic drug monitoring.
Acknowledgements
CE
This study was supported by grants from the research council of Southeast Sweden (FORSS) and
AC
12
ACCEPTED MANUSCRIPT
References
T
concentration by bioassay, a valid method for therapeutic drug monitoring for clinical laboratories.
IP
Antimicrob. Agents Chemother. 57, 3437-3440 (2013).
R
Candejas-Bueno, E., Forastiero, A., Rodriguez-Tudela, J.L., Cuenca-Estrella, M. and Gomez-Lopez,
A. HPLC/UV or bioassay: two valid methods for posaconazole quantification in human serum
SC
samples. Clin. Microbiol. Infect. 18, 1229-1235 (2012).
Carlier, M. et al. Assays for therapeutic drug monitoring of -lactam antibiotics: A structured
NU
Review. Int. J. Antimicrob. Agents. 46, 367-375 (2015).
Cotar A-I. et al. Screening of Molecular Virulence Markers in Staphylococcus aureus and
Pseudomonas aeruginosa Strains Isolated from Clinical Infections. Int. J. Mol. Sci. 11, 5273-5291
MA
(2010).
Dafale, N.A., Semwal, U.P., Agarwal., Sharma, P. and Singh, G.N. Quantification of ceftriaxone
sodium in pharmaceutical preparations by a new validated microbiological assay. Anal. Methods. 4,
D
2490-2498 (2012).
TE
Erickson, R.C., Hildebrand, A.R., Hoffman, P.F. and Gibson C.B. A sensitive bioassay for
teicoplanin in serum in the presence or absence of other antibiotics. Diagn. Microbiol. Infect. Dis.
AC
13
ACCEPTED MANUSCRIPT
Mendez, A.S.L., Weisheimer, V., Oppe, T.P., Steppe, M. and Schapoval, E.E.S. Microbiological
assay for determination of meropenem in pharmaceutical dosage form. J. Pharm. Biomed. Anal. 37,
649-653 (2005).
Mller, D. M. and Rentsch, K.M. Therapeutic drug monitoring by LC-MS-MS with special focus on
anti-infective drugs. Anal. Bioanal. Chem. 398, 2573-2594 (2010).
T
Roberts, J.A. et al. DALI: Defining antibiotic levels in intensive care unit patients: Are current levels
IP
of -lactam antibiotic doses sufficient for critically ill patients? Clin. Infec. Dis. 58, 1072-1083
(2014).
R
Sargent, M. (Ed.). Guide to achieving reliable quantitative LC-MS measurements 1st Edition. RSC
SC
Analytical Methods Committee. (2013).
Suthakaran, C. and Adithan, C. Therapeutic drug monitoring concepts, methodology, clinical
NU
application and limitations. Health. Admin. 19, 22-26 (2006).
Taccone, F.S. et al,. Insufficient -lactam concentrations in the early phase of severe sepsis and
septic shock. Critical Care. 14, R126 (2010).
MA
White, R.L., Burgess, D.S., Manduru, M. and Bosso, J.A. Comparison of three different in vitro
methods of detecting synergy: Time-kill, checkerboard and E test. Antimicrob. Agents Chemother.
40, 1914-1918 (1996).
D
Wiegand, I., Hilpert, K. and Hancock, R.E.W. Agar and broth dilution methods to determine the
TE
14
ACCEPTED MANUSCRIPT
T
concentrations attained throughout the study. Total tests performed (n), along with the number of occasions where the observed MICControl were within the
IP
acceptable EUCAST MIC range.
CR
EUCAST EUCAST Agreement
US
Bacterial > +1 MIC Within 1 MIC <(-1) MIC with
Antibiotic ATCC target MIC MIC range n
strain step step step
N
(mg/l) (mg/l) EUCAST
MA
S. aureus 29213 2.0 14
CTX 203 5 191 7 94.1%
E. coli 25922 0.125 0.064 0.25
D
P. aeruginosa 27853 0.5 0.25 1
TE
MER 221 3 216 2 97.7%
E. coli 25922 0.016 0.008 0.032
P
PIP E. coli 25922 2.0 14 235 7 228 0 97.0%
CE
AC
15
ACCEPTED MANUSCRIPT
15
M ic r o b io lo g ic a l A s s a y
T
10
R IP
SC
5
0
0 5 10 NU 15 20 25
MA
L C - M S /M S
Figure 1A. Correlation plot of samples from ICU patients treated with cefotaxime analysed by LC-MS
and the microbiological assay respectively (p<0.0001. r= 0.86. n = 31). Squares = Samples analysed by
D
diluting bacteria in normal human serum. Circles = Samples analysed by diluting bacteria in Mueller-
TE
Hinton broth.
P
CE
AC
16
ACCEPTED MANUSCRIPT
20
M ic r o b io lo g ic a l A s s a y
15
T
IP
10
R
SC
5
0
0 10 NU
20 30
MA
L C - M S /M S
Figure 1B. Correlation plot of samples from ICU patients treated with meropenem analysed by LC-
MS and the microbiological assay respectively (p< 0.0001. r= 0.96 n = 18). Squares = Samples
D
analysed by diluting bacteria in normal human serum. Circles = Samples analysed by diluting bacteria
TE
in Mueller-Hinton broth.
P
CE
AC
17
P ip e r a c illin
ACCEPTED MANUSCRIPT
M ic r o b io lo g ic a l A s s a y 200
150
T
IP
100
R
SC
50
0 NU
MA
0 50 100 150 200
L C - M S /M S
D
Figure 1C. Correlation plot of samples from ICU patients treated with piperacillin analysed by LC-
MS and the microbiological assay respectively (p< 0.0001. r= 0.88. n = 39). Squares = Samples
TE
analysed by diluting bacteria in normal human serum. Circles = Samples analysed by diluting bacteria
in Mueller-Hinton broth.
P
CE
AC
18
ACCEPTED MANUSCRIPT
Highlights
T
R IP
SC
NU
MA
D
P TE
CE
AC
19