DOI 10.1515/cclm-2013-0142
Clin Chem Lab Med 2013; 51(9): 1761–1769
Katrin König*, Uwe Kobold, Gerhard Fink, Andreas Leinenbach, Thomas Dülffer,
Roland Thiele, Johannes Zander and Michael Vogeser
Quantification of vancomycin in human serum by
LC-MS/MS
Abstract
Background: The aim of our work was to develop and
validate a reliable LC-MS/MS-based measurement proce-
dure for the quantification of vancomycin in serum, to be
applied in the context of efforts to standardize and harmo-
nize therapeutic drug monitoring of this compound using
routine assays.
Methods: Sample preparation was based on protein pre-
cipitation followed by ultrafiltration. In order to minimize
differential modulation of ionization by matrix constitu-
ents extended chromatographic separation was applied
leading to a retention time of 9.8 min for the analyte.
Measurement was done by HPLC-ESI-MS/MS. For internal
standardization the derivative vancomycin-glycin (ISTD)
prepared by chemical synthesis was used, HPLC condi-
tions ensured coelution of ISTD with the analyte.
Results: In a bi-center validation total CVs of < 4% were
observed for quality control material ranging from
5.3mg/L to 79.4 mg/L; accuracy was ± 4%. No relevant ion
suppression was observed. Comparative measurement of
aliquots from 70 samples at the two validation sites dem-
onstrated close agreement.
Conclusions: Employing a closely related homologue
molecule for internal standardization and the use of MS/
MS following highly efficient sample pre-fractionation by
HPLC, the method described here can be considered to
offer the highest level of analytical reliability realized so
far for the quantification of vancomycin in human serum.
Thus, the method is suitable to be used in a comprehen-
sive reference measurement system for vancomycin.
Keywords: LC-MS/MS; liquid chromatography; mass spec-
trometry; serum; vancomycin.
*Corresponding author: Katrin König, Institute of Laboratory
Medicine, Hospital of the University of Munich, Marchioninistrasse
15, 81375 Munich, Germany, Phone: +49 89 70953221,
E-mail: katrin.koenig@med.uni-muenchen.de
Uwe Kobold, Gerhard Fink, Andreas Leinenbach, Thomas Dülffer
and Roland Thiele: Roche Diagnostics GmbH, Penzberg, Germany
Michael Vogeser and Johannes Zander: Institute of Laboratory
Medicine, Hospital of the University of Munich, Munich, Germany
Introduction
The glycopeptide compound vancomycin is one of the
most widely used antimicrobial agents for the treatment
of serious gram-positive infections including methicillin-
resistant Staphylococcus aureus (MRSA) [1]. Vancomycin is
also among the highest volume target analytes in thera-
peutic drug monitoring (TDM) for many years now [1–3].
TDM of vancomycin mainly aims to balance therapeutic
efficacy against the risk of nephrotoxicity. It has been
shown in several studies – applying various analytical
methods – that high vancomycin trough levels are asso-
ciated with the incidence and extend of nephrotoxicity
[4,5]. However, low trough levels of vancomycin may lead
to increased occurrence of resistant strains of S. aureus
and failure of treatment in complicated infections. Based
on these issues the American Society of Health-System
Pharmacists, the Infectious Diseases Society of America,
and the Society of Infectious Diseases Pharmacists have
published expert panel recommendations for vancomycin
TDM, recommending trough serum concentrations of van-
comycin of 15–20 mg/L in complicated infections [1].
These recommendations, however, do not take into
consideration that routine serum vancomycin quanti-
fication by commercial immunoassays is still lacking
between-method standardization and substantial method
bias can be found [6]. In the proficiency testing scheme of
the German Association of Clinical Chemistry and Labo-
ratory Medicine [DGKL; Referenzinstitut für Bioanalytik
(RfB), Bonn, Germany] at present 14 tests are monitored.
The bias between the lowest reading vancomycin test
and the highest reading test is continuously found in the
range of 40% in this external quality assessment program.
Considering the clear-cut vancomycin target concentra-
tion ranges, it is likely that different clinical dosing deci-
sions are made today in a substantial number of patients
depending on the assay which is used in an individual
institution. Notably, these consensus target concentra-
tions ranges cannot be traced back conclusively to a
defined analytical method.
From these considerations it has to be concluded that
improved harmonization and standardization of serum
Bereitgestellt von | Bayerische Staatsbibliothek
Angemeldet
Heruntergeladen am | 05.11.15 11:17
1762
König etal.: Quantification of vancomycin in human serum by LC-MS/MS
vancomycin measurement is warranted. A future compre-
hensive reference measurement system for vancomycin
measurement has to include, on the one hand, reliable
reference materials, but on the other hand, a robust and
reliable method for the specification of working calibration
materials, and proficiency testing materials, as well as for
the evaluation of routine immunometric tests referring to
large reference serum panels. The aim of our work was to
develop and to validate such a candidate reference method.
Due to the rather high molecular weight and the
limited thermal stability of vancomycin, LC-MS/MS was
the most promising technology for this aim. Indeed,
several LC-MS/MS methods for the quantification of
serum vancomycin concentrations have been described
previously [7–9]. These methods, however, rely on inter-
nal standard compounds with are structurally not related
to the target analyze (teicoplanin, atenolol).
LC-MS/MS-based reference methods [10] usually
involve stable isotope labeled compounds for internal
standardization. Since vancomycin is a bio-product, pro-
duction of a respective material is hardly possible. Conse-
quently we designed a suited internal standard compound
by stable chemical derivatization of vancomycin which
showed almost identical behavior during sample prepa-
ration and HPLC-MS/MS separation. The molecule was
modified only marginal by introduction of a small func-
tional group not changing the polarity of the molecule
leading to a different molecular mass.
The intended use of this method protocol described
herein is the specification of serum-based samples within
the calibration range of the method in both an industrial
and research setting. Consequently, high sample through-
put did not have major priority in the development of this
method, while robustness and reproducibility in differ-
ent instrument settings was an essential goal. Thus, we
decided to apply highly efficient chromatographic frac-
tionation in order to minimize matrix effects, and to apply
an innovative bi-centric validation protocol.
Materials and methods
Chemicals and reagents
Vancomycin-hydrochloride pure substance was purchased as a USP
reference standard from U.S. Pharmacopeia (USP Rockville MD, USA;
LOT M0H006; CAS 1404–93–9; molecular weight 1485.71; molecular
formula C
66
H
75
Cl
2
N
9
O
24
.HCl; vials containing 99,300 µg vancomycin
activity; Figure 1A). USP-based routine human serum calibrators for
immunoassays (Cal A, 5 mg/L; Cal B, 40 mg/L; Cal C, 80 mg/L) were
purchased from Roche Diagnostics (Mannheim, Germany). The latter
samples were used for investigation of accuracy and reproducibility.
A
B
C
Figure 1Synthesis of vancomycin-glycine.
(A) Vancomycin (molecular weight: 1449.25; molecular formula:
C
66
H
75
Cl
2
N
9
O
24
). The only carbon acid function accessible for forma-
tion of an ester derivative is indicated. (B) Peptide coupling of van-
comycin to vancomycin-glycine-methyl ester. (C) Saponi fication of
vancomycin-glycine-methyl ester to vancomycin-glycine (molecular
weight: 1506.34; molecular formula: C
68
H
78
Cl
2
N
10
O
25
).
Methanol and water (both HPLC-grade) for chromatography
were from J.T. Baker (Griesheim, Germany). Formic acid and trichlo-
roacetic acid were from Merck (Darmstadt, Germany).
For synthesis of a derivative of vancomycin to be used as an
internal standard, vancomycin-hydrochloride and glycine methyl
ester hydrochloride were purchased from Sigma-Aldrich (Steinheim,
Germany). O-(Benzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium hex-
auorophosphate (HBTU), Dimethyl sulfoxide (DMSO), N,N-Dimeth-
ylformamide (DMF), diisopropylethylamine, acetic acid, sodium car-
bonate, water and acetonitrile were from J.T. Baker.
Synthesis of vancomycin-glycine
Peptide coupling
Vancomycin-hydrochloride (200 mg; 135 µmol), glycine methyl ester
hydrochloride (33.8 mg; 269 µmol) and HBTU (76.5 mg; 202 µmol)
were dissolved in DMSO (2 mL) and DMF (0.66 mL). The resulting
Bereitgestellt von | Bayerische Staatsbibliothek
Angemeldet
Heruntergeladen am | 05.11.15 11:17
König etal.: Quantification of vancomycin in human serum by LC-MS/MS
1763
solution was cooled to 0°C, and diisopropylethylamine (117 µL;
673 µmol) was added. Aer stirring at room temperature for 2.5 h,
the mixture was diluted by acetic acid (0.5 mL), water (2 mL) and ace-
tonitrile (0.5 mL). Using preparative reverse phase HPLC (C18; Vydac
218TP152050; 5 × 25 cm) the reaction product vancomycin-glycine-
methyl ester (Vancomycin-Gly-OMe; Figure 1B) was puried and sub-
sequently dried to a white powder (approx. 150 mg).
Saponification of vancomycin-glycine-methyl ester
Vancomycin-Gly-OMe (75 mg; 49.3 µmol) was added to sodium car-
bonate buer (2%; 5 mL; pH 10.0) at room temperature. The reac-
tion mixture was stirred for 6h and then acidied with acetic acid
(0.5 mL). The main reaction product vancomycin-glycine (Figure 1C)
was again puried by preparative reverse phase HPLC. Aer drying a
white powder was obtained.
For preparative HPLC separation, a Dionex P580P HPG HPLC
Binary High-pressure Gradient Pump (Thermo Scientic, Sunnyvale,
USA), a Rheodyne switching valve (IDEX, Rohnert Park, USA), a
Gynkotek SP-6 detector (Thermo Scientic) and a LKB 2211 Superrac
fraction collector (AIE, Haverhill, USA) were used. As column a Vydac
C 18 (218TP152050; 5 × 25 cm) (Grace, Worms, Germany) was used.
The mobile phase consisted of Eluent A [aqueous triuoroacetic acid
(0.1% v/v)] and Eluent B [80% acetonitrile/20% water containing tri-
uoroacetic acid (0.1% v/v)], delivered with a ow rate of 35 mL/min
during a run time of 120 min. The gradient elution program was as
follows: 100% eluent A for 100 min; increase to 25% eluent B over
3 min, aerwards an increase up to 100% eluent B hold for 7 min.
Separation was monitored at 226 nm. The uniform peak fraction elut-
ing around 70min was collected (Figure 2). The solvent was removed
under low pressure and the residue was brought to dryness under
vacuum to obtain approximately 40mg of a white solid.
The nal product of this procedure was analyzed by ESI-TOF-MS
in the positive mode; this oered the following fragmentation
patterns: m/z 1507 (M+H); 1364 [(Vanc-143)
+
]; 754 [(M+2H)
2+
], which
is evidentiary for Vancomycin-Gly; (Vanc-143)
+
corresponds to the
loss of the 4-amino-4-methyl-5-hydroxy-6-methyl-glucose residue
( Figure1A). Since the molecular structure of the vancomycin mole-
cule oers only one carboxylic acid function which is available for
reaction to an ester (Figure 1A), such uniform molecular structure of
vancomycin-glycine can be expected.
Preparation of stock solutions, calibrators
and quality control samples
The dry pure substance of vancomycin weighted into vials by USP
was dissolved in 100 mL of HPLC-grade water according USP han-
dling instructions, leading to a stock solution containing 993 mg/L
vancomycin based on USP certicate. An internal standard working
solution (5 mg/L) was prepared by dissolving vancomycin-glycine in
HPLC-grade water.
Drug-free human serum was spiked with vancomycin stock solu-
tion to yield the following ve calibrator concentrations: 1.06 mg/L;
21.1 mg/L; 42.2 mg/L; 63.3 mg/L; and 84.4 mg/L. Calibrator 0 was
drug free human serum. Calibrator samples were stored at −20°C, the
internal standard solution at +4°C during the study.
Quality control samples (QC) in ve concentration levels were
prepared. QC 1–3 were mixtures of le-over patients’ samples found
in dierent concentration ranges in routine analyses (for QC 1, sam-
ples between 1 and 9 mg/L; for QC 2, samples between 9 and 13
mg/L; and for QC 3 samples > 13 mg/L). QCs 1–3 were used to assess
the reproducibility of the method. QC 4 and QC 5 were prepared by
spiking drug-free human serum with a vancomycin working solu-
tion to a concentration of 40 mg/L, and 80 mg/L, respectively. This
mV
%D:0.0%
%C:0.0%
700
850
TB404_7143_GF #3 [modified by finkgd] F1884
80%ACN+20%H2O+0.1% TFA:0.0%
Flow:35.000mL/min
UV1
100.0
25.0
600
500
400
300
200
100
-30
0102030405060708090 100 110 120
0.000
min
Figure 2Preparative chromatography applied to isolate vancomycin-glycine eluting after approximately 70 min.
Bereitgestellt von | Bayerische Staatsbibliothek
Angemeldet
Heruntergeladen am | 05.11.15 11:17
1764
König etal.: Quantification of vancomycin in human serum by LC-MS/MS
vancomycin solution was prepared separately from the solution used
for the preparation of calibrator samples. All QC samples were ali-
quoted and stored at −20°C until analysis on either validation site.
High-performance liquid chromatography
conditions
At laboratory site 1 (Munich) a Waters Alliance 2795, at site 2
( Penzberg) a Waters Acquity LC system was used. As analytical
column a Fortis C8 (100 mm × 2.1 mm, 3 µm) (dichrom GmbH, Marl,
Germany) was used. The column temperature was 40°C. The injection
volume was 20 µL. The mobile phase consisted of two solvents; Eluent
A: aqueous formic acid (0.1% v/v); Eluent B: methanol containing
0.1% formic acid (0.1% v/v). Run time was 21min with a ow rate of
0.3 mL/min and an gradient elution program as follows: 100% A for
3 min; linear increase to 30% B over 7 min, hold for 2 min; aerwards
a linear increase up to 80% B within 1 min, hold for 2 min; return to
the initial condition within 1min and re-equilibration for 5min. Via
a post-column switching valve, the HPLC eluate was directed into the
mass spectrometer between 4.0 and 10.5 min aer injection; during
the residual run time the eluent was diverged into waste.
Mass spectrometric conditions
Mass spectrometric analysis was performed using Waters Quattro
Micro instruments in the positive ionization mode on both laboratory
sites. The following settings were applied: Capillary voltage, 3.2 kV;
cone voltage, 20.0 V; collision energy, 12 eV; source temperature,
120°C; and desolvation temperature 480°C. The dwell time was 0.2ms;
inter-channel delay, 0.02 ms; and inter-scan delay 0.02 ms. Mass spec-
trometric data were acquired from 4.5 to 12.0min aer injection.
A collision-induced product ion scan of vancomycin is shown in
Figure 3. For vancomycin the mass transition 725 > 1306 was recorded,
and for vancomycin-glycine 753 > 1362. This corresponds to the dou-
bly charged protonated molecules as the precursor ions, and singly
charged fragments as the product ions – leading to higher m/z values
of the product ions compared to the precursor ions. Mass resolution
was tuned to obtain a mass signal width of 1 Da at 50% height of the
product ion signal.
Sample preparation
Seventy-ve µL of internal standard working solution and 75 µL of the
calibrator sample or of unknown serum sample were pipetted into
1.5 mL polypropylene microcentrifuge tubes. For equilibration the
solution was shaken at room temperature for 30 min. Subsequently,
300 µL of trichloroacetic acid (15%) were added. Protein precipita-
tion was achieved by shaking at room temperature for 5 min. The
tubes were centrifuged at 14,000 g for 10 min. The clear supernatant
was pipetted into Amicon Ultra-0.5mL 10k ultra-ltration devices –
(Millipore, Billerica, USA; 10,000 nominal molecular weight cut-o).
The devices were placed into a centrifuge and ltration was achieved
by centrifugal force for 45min at 7000 g. The ltrate (approx. 450 µL)
was transferred into HPLC vials with small-volume inserts.
Five-point calibration was applied in a concentration range
from 1.06 to 84.4 mg/L. For quantication the Waters QuanLynx so-
m/z
500600 700800 900100011001200130014001500
%
0
100
VANCO DAUGHTERSCAN 17 (0.314) Sm (SG, 2x2.00); Cm (4:28) Daughters of 725ES+
1.35e5
1306.16
1142.87
724.81
571.94
1276.32
1331.83
Figure 3Production scan of two-fold charged vancomycin.
Bereitgestellt von | Bayerische Staatsbibliothek
Angemeldet
Heruntergeladen am | 05.11.15 11:17
König etal.: Quantification of vancomycin in human serum by LC-MS/MS
1765
ware module was used with the following setting: Polynome type,
linear; origin, include; weighting function, 1/x; axis transformation,
none; smoothing method, mean; smoothing width, 1; smoothing it-
erations, 1.
Method validation
The method was validated in a bi-centric protocol: Institute of Labo-
ratory Medicine, Hospital of the University of Munich, Germany, site
1; and Roche Diagnostics, Penzberg, Germany, site 2.
Prior to each analytical run, a system suitability screening test
was performed; for this the preparation of calibrator 1 (1 µg/mL) was
injected. A signal-to-noise ratio of > 10:1 for the MRM-trace of vanco-
mycin was dened as mandatory for a subsequent analytical run.
In order to test the specicity of the method, 10 leover clinical
serum samples from intensive care patients not treated with vanco-
mycin were used. These samples were processed and analyzed with-
out addition of the internal standard solution in order to verify the
absence of peaks within the retention time window of the analyte or
the internal standard, respectively.
In order to characterize the potential impact of residues of the
sample matrix on the ionization of analyte and internal standard,
two investigations were performed. In the post-column-infusion set-
ting [11], a pure solution of vancomycin (1 mg/L) was infused with a
syringe pump at constant rate using a T-piece into the column eu-
ent, in order to generate a MS/MS background signal. Upon injection
of processed serum samples, potential modulation of the background
signal was monitored. In a second experiment, extracted serum from
three patients not treated with vancomycin was spiked with a solu-
tion of vancomycin to a concentration of 21 mg/L each. For compari-
son, a neat sample in water was spiked in the same way in triplicate.
Both sets of samples were analyzed with the MS/MS method and
peak areas were recorded in order to estimate potential eects of the
serum-derived sample matrix on ionization of vancomycin, as matrix
eect (%) according to Matuszewski etal. [12].
Accuracy of the method was tested by analyzing three external
reference calibrator samples (USP Calibrators, Roche) and the qual-
ity control samples QC 4 and QC 5 which were prepared on valida-
tion site 2. Each sample was analyzed in triplicate on both validation
sites.
Imprecision of the method was studied by analyzing aliquots of
the QC samples 1–5 in two series of six-fold determination on both
study sites, leading to a total of 24 results for each sample.
To roughly characterize a lower limit of detection and the per-
formance of the method implementation in a concentration range be-
low the lowest calibrator sample, a spiked sample with a vancomycin
concentration of 0.1 µg/mL was injected in triplicate on both study
sites and the signal-to-noise ratio was assessed.
In order to test the stability of processed samples, extracts of
three patients’ samples were stored at +8°C and −20°C. These sam-
ples were re-quantied in analytical series aer 1 day, 2 days, and
1 week. These results were compared with the results found in the
initial analytical run.
Le-over serum from 70 patients’ samples send to the Institute
of Laboratory Medicine for clinically indicated vancomycin measure-
ment were used to study the agreement between analyses performed
with the described LC-MS/MS method on both validation sites. The
samples were recruited consecutively without any selection, thus re-
ecting the typical distribution of concentrations found in a tertiary
care hospital. Aer anonymization, two aliquots of these samples
were prepared and stored at −20°C until analysis on either study site
within 4 weeks. This procedure was approved by the Institutional
Review Board.
Results
A representative LC-MS/MS chromatogram is shown in
Figure 4. Analyte and the internal standard compound
vancomycin-glycine co-eluted as requested. Fifty MS/
MS-data points were acquired over both chromato-
graphic peaks. Minor signals from isobaric compounds
were observed eluting slightly beforehand and follow-
ing the peaks. Their pattern was found constant in both
calibrator samples and patients’ samples on both study
sites, and peak integration was performed consistently.
The pattern did not change with the storage time of
samples.
In all analytical series the regression coefficient r
2
was ≥ 0.99 for all calibration runs over the concentration
range from 1.06 to 84.41 mg/L. The slope of the calibration
line was 0.25 ( ± 5%).
The signal-to-noise ratio observed in the system suit-
ability screening test (injection of the lowest calibrator
sample) was > 400:1 in all analytical series on study site 1,
and > 100:1 on site 2.
The handling of the method was found convenient,
including the ultrafiltration step. The entire instrumental
setting was robust throughout the study period on both
sites, in particular regarding HPLC back-pressure and ion-
ization yield.
The analysis of 10 serum samples from patients which
were not treated with vancomycin proved the specificity of
the method; no peak signals in retention time windows of
vancomycin or vancomycin-glycin were observed.
Figure 5 displays the results of the post-column infu-
sion experiment which was performed to characterize the
impact of serum-derived sample materials on the ioniza-
tion yield of vancomycin. The figure shows an overlay of
the signal pattern generated by injection of extract from
a vancomycin-free sample, and from a patient’s sample.
During the time period in which the chromatographic
eluate was transferred to the MS/MS system a con-
stant signal of the continuously infused analyte can be
observed. There was no obvious drop or increase in ion
yield at the retention time of vancomycin. In the spiking
experiment, a matrix effect of –9% according to Matusze-
wski etal. [12] was observed.
Table 1 displays the analytical accuracy realized for
the analysis of standard and quality control materials;
Bereitgestellt von | Bayerische Staatsbibliothek
Angemeldet
Heruntergeladen am | 05.11.15 11:17
