RAPID COMMUNICATIONS IN MASS SPECTROMETRY

Rapid Commun. Mass Spectrom. 14, 14041409 (2000)

Confirmatory assay for the simultaneous

detection of penicillins and cephalosporins in milk
using liquid chromatography/tandem mass
spectrometry
Els Daeseleire*, Hendrik De Ruyck and Roland Van Renterghem
Ministry of Small Enterprises, Traders and Agriculture, Agricultural Research Centre-Ghent, Animal Product Quality and
Transformation Technology, Brusselsesteenweg 370, 9090 Melle, Belgium

SPONSOR REFEREE: Dr. Michael OKeefe, The National Food Centre, Dublin, Ireland

A high-performance liquid chromatography/tandem mass spectrometry (LC/MS/MS) method for the

detection of residues of penicillins and cephalosporins in milk has been developed. After a simple extraction
with acetonitrile, the extract was directly injected into the LC/MS/MS system on a C18 column. A gradient
consisting of acetonitrile and water, each containing 0.1% formic acid, was applied. The abundant parent
ions [M H] produced by positive electrospray ionisation were selected for fragmentation with argon. For
each compound at least one fragment was recorded with multiple reaction monitoring. The limits of
detection ranged from 1.5 to 25 mg/kg and the limits of quantification ranged from 4 to 50 mg/kg. Recoveries
were examined at three levels (MRL, 0.5  MRL, 2  MRL) and ranged from 57 to 88%. The coefficients of
variation obtained for the repeatability experiments were in agreement with those specified by the Horwitz
equation. Linearity was checked by injecting extracts of samples spiked with increasing amounts of the
different standards ranging from 0 to 150 mg/kg. The advantage of this method over existing methods is the
very simple sample pre-treatment which makes the method very suitable for routine analysis. Copyright
# 2000 John Wiley & Sons, Ltd.
Received 7 June 2000; Accepted 8 June 2000

b-Lactam antibiotics are very widely used in veterinary
medicine for the treatment and prevention of diseases. This
use can result in the presence of residues in milk and edible
tissues which can lead to problems in fermentation
processes and to health problems for individuals who are
hypersensitive to b-lactams.1 This is especially the case if
proper withdrawal periods are not respected. The develop-
ment of antibiotic resistant bacteria should not be under-
estimated. To protect the consumer, MRL-values were laid
down in EEC regulation 2377/90,2 and subsequent mod-
ifications. For milk these values are 4 mg/kg for benzyl-
penicillin, ampicillin, amoxicillin and penethamate, 10 mg/
kg for cephapirin and cefalonium, 20 mg/kg for cefquinome,
30 mg/kg for dicloxacillin, cloxacillin, oxacillin and nafcil-
lin, 50 mg/kg for cefazolin and cefoperazon, 100 mg/kg for
ceftiofur and cefalexin and 125 mg/kg for cefacetril (as of
February 2000).
In practice, screening is usually performed by micro-
biological and immunological methods. In this way it is
possible to determine, for positive samples, the group to
which the residue belongs. Confirmation and specific
substance detection, however, has to be performed by an
*Correspondence to: E. Daeseleire, Ministry of Small Enterprises,
Traders and Agriculture, Agricultural Research Centre-Ghent, Animal
Product Quality and Transformation Technology, Brusselsesteenweg
370, 9090 Melle, Belgium.
E-mail: e.daeseleire@clo.fgov.be
independent physicochemical technique. Another drawback
of the microbiological methods is that the limits of detection
are for a lot of compounds higher than the MRL-value. Very
many HPLC methods have been developed during the past
15 years; however, the number of papers dealing with
methods capable of determining both penicillins and
cephalosporins is limited.39 Therefore a rapid, simple and
reliable LC/MS/MS method, which can be used on a routine
basis, was developed for the simultaneous detection of 11
b-lactam antibiotics (benzylpenicillin, ampicillin, amoxi-
cillin, oxacillin, cloxacillin, dicloxacillin, nafcillin, ceftio-
fur, cefazolin, cefalexin and cephapirin) in milk.
EXPERIMENTAL
Reagents and chemicals
Amoxicillin, ampicillin sodium salt, cefazolin sodium salt,
cephalexin hydrate, cephapirin sodium salt, cloxacillin
sodium salt, dicloxacillin sodium salt, nafcillin, oxacillin
sodium salt and penicillin-G sodium salt were purchased
from Sigma (St. Louis, MO, USA). Ceftiofur was a gift from
Pharmacia and Upjohn (Kalamazoo, MI, USA). The
chemical structures of the compounds are shown in Fig. 1.
Acetonitrile was gradient grade Lichrosolv from Merck
(Darmstadt, Germany). Formic acid was from UCB
(Leuven, Belgium). Water was HPLC grade (generated by
an ELGA purification system). Filters for filtration of the

Copyright # 2000 John Wiley & Sons, Ltd.

 DETECTION OF PENICILLINS AND CEPHALOSPORINS IN MILK 1405

Figure 1. Chemical structures of the examined penicillins and cephalosporins.

extract were from Millipore (Millex GV, 0.22 mm). tC18 0.25 mL/min. A split device was incorporated and was
extraction columns (Sep-Pak, 500 mg, 6 ml) were from regulated to a split ratio of 1:1.
Waters (Milford, MA, USA). The Quattro LCZ mass spectrometer was operated in the
Standard stock solutions (1 mg/mL) were prepared in ESI-MS/MS positive ion mode. High-purity nitrogen was
water/acetonitrile (50:50, v/v) and stored at 4 C. Fresh used as drying gas and as ESI nebulising gas. The MS
solutions were prepared every month. A dilution of 1 ng/mL parameters were optimised using standard solutions of 1 ng/
in water/acetonitrile (50:50, v/v), containing 0.1% formic mL in water/acetonitrile (50:50, v/v) containing 0.1% formic
acid, was used for tuning and optimisation of the instrument. acid. Argon was used as collision gas to obtain daughter
For spiking the milk, fresh dilutions (10, 1 and 0.1 ng/mL) ions. A summary of the cone voltages, collision energies,
were made in water. and parent and daughter ions of the compounds is presented
in Table 1. The source block and desolvation temperature
were set at 80 and 250 C, respectively.
Apparatus
The HPLC system consisted of a Kontron system (ternary
pump 325, vacuum degasser degasys DG 1310, autosampler Table 1. Optimised ESI() mass spectrometric conditions for
infusion of standard solutions at 1 ng/mL (underlined
465) (Biotech Instruments, Milan, Italy) coupled to a fragment ions were followed during SRM)
Micromass (Altrincham, Cheshire, UK) Quattro LC (triple
Parent ion Daughter ions Collision energy Cone voltage
quadrupole) equipped with a Z-spray system. The system Compound (m/z) (m/z) (eV) (V)
was controlled by version 3.3 of the Masslynx software. Cefalexin 348 158, 174 10 24
Chromatography was performed on an Alltima C18 column Ampicillin 350 106, 192, 160 15 24
(5 mm, 150 mm 2.1 mm i.d.) protected by a guard column Amoxicillin 366 208, 349, 114 10 20
containing the same material. A gradient was applied with Cephapirin 424 292, 152, 320 20 24
water (A) and acetonitrile (B), each containing 0.1% formic Cefazolin 455 323, 156 15 24
acid. The gradient conditions were as follows: From 0 Ceftiofur 524 241, 285, 210 20 34
0.5 min, hold 100% A; ramp over 0.1 min to 55% A and Penicillin G 335 160, 176 20 23
45% B; ramp over 7.9 min to 35% A and 65% B; ramp over Oxacillin 402 160, 243 15 24
Cloxacillin 436 160, 277 12 24
0.1 min to 100% B; hold for 1 min; ramp over 0.2 min to
Nafcillin 415 199, 256 28 24
100% A. Hold 100% A for 7.2 min to reequilibrate the Dicloxacillin 470 160, 311 15 23
system. The total run time was 17 min. The flow was

Copyright # 2000 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 14, 14041409 (2000)
1406 DETECTION OF PENICILLINS AND CEPHALOSPORINS IN MILK

Figure 2. Chromatograms of a blank raw milk sample spiked with 2 mg/kg of penicillin G,
oxacillin, dicloxacillin, nafcillin and ampicillin, 5 mg/kg of ceftiofur and cloxacillin, 10 mg/kg of
cephalexin, cephapirin and amoxicillin and 15 mg/kg of cefazolin.

METHOD standard (20 mg/kg) and added after extraction/clean-up of

Sample clean-up
Portions of milk (2g) were mixed with 4 mL of acetonitrile
for deproteinization. After vortex-mixing the sample was
centrifuged for 10 min. at 3500 rpm. After filtration, 20 mL
of the supernatant were injected into the LC/MS/MS
system.
Validation study
Recovery of the extraction procedure was determined at
three different levels (MRL, 0.5  MRL and 2  MRL)
using spiked milk samples. Nafcillin was used as internal
the samples. The ratio of the area of a compound to the area
of the internal standard obtained for samples spiked before
clean-up were divided by the ratio obtained for a blank
matrix that was spiked after clean-up and multiplied by 100
to obtain the percentage recovery. For the recovery of
nafcillin, ceftiofur (20 mg/kg) was used as internal standard.
The limits of detection (LODs) were those concentrations
of analyte which yielded a signal-to-noise ratio (S/N) of at
least 3:117 and the limits of quantification (LOQs) were
chosen as those concentrations which yielded a S/N of at
least 6:1. Blank milk samples spiked at different levels
(from 0 to 50 mg/kg) were used to determine these values.
Linearity was checked by injecting extracts of samples

Rapid Commun. Mass Spectrom. 14, 14041409 (2000) Copyright # 2000 John Wiley & Sons, Ltd.
 DETECTION OF PENICILLINS AND CEPHALOSPORINS IN MILK 1407

spiked with increasing amounts of the different standards Table 2. Recovery experiments at three different levels in farm
ranging from 0 to 150 mg/kg and a constant amount of milk (n = 6, each level)
internal standard. Mean recovery (%)  SD
Repeatability experiments were carried out at the MRL MRL
Compound (mg/kg) 0.5  MRL MRL 2  MRL
levels. Specificity was checked by analysing blank milk
samples. Penicillin G 4 85  14 86  6 85  8
Ampicillin 4 59  14 78  5 77  4
Amoxicillin 4 Only performed at 20 mg/kg : 57  12
RESULTS AND DISCUSSION Cephapirin 10 65  8 78  7 96  11
Cloxacillin 30 85  6 82  5 86  5
Tuning of the LC/MS/MS instrument was performed with Dicloxacillin 30 88  6 83  7 86  6
solutions containing 1 ng/mL of the compounds. For Oxacillin 30 84  5 86  4 87  2
electrospray, the best results were obtained in the positive Nafcillin 30 86  2 87  8 85  2
ionisation mode. The negative ion mode of operation was Cefazolin 50 67  4 63  3 83  5
several times less sensitive for all compounds tested, as Ceftiofur 100 88  3 86  4 88  6
found also by other authors.3,10 The addition of 0.1% formic Cefalexin 100 80  9 78  5 81  5
acid to the standard solution increased the abundance of the
[MH] response. Too high a concentration of formic acid
(1%) gave abundances which were lower then those
obtained without adding acid. For each compound the Normally the baffle and sample cone were cleaned every
[MH] ion was determined and dissociation with argon of two or three weeks, which took about 1 hour.
the molecular ion was induced. The MS system was tuned to Chromatography was performed on a reversed phase C18
obtain the optimal conditions for generating the daughter column. Complete separation was not accomplished but this
ions. The MS parameters where almost all parent ions are was not necessary due to the very specific detection of
converted to daughter ions are shown in Table 1. The multiple reaction monitoring (MRM). A chromatogram
fragmentation patterns and the explanation for the frag- obtained in MRM mode of a spiked raw milk sample
ments obtained of penicillins and cephalosporins have been indicating the retention times is shown in Fig. 2.
discussed already by other authors36,1015 and are beyond In practice, for an unknown sample, a first injection of the
the scope of this work. However, it can be said that the sample was performed using a MS method in which the
fragments obtained in the triple quadrupole of the Micro- transition from the [M H] ion to one daughter ion was
mass Quattro LCZ instrument are comparable with the followed (underlined in Table 1). This technique is called
fragments obtained by Heller and Ngoh using a Finnigan selected reaction monitoring (SRM). The instrument con-
tandem ion trap instrument (LCQ).9 sists of an ion source, three quadrupole mass filters and an
Several clean-up procedures for the milk samples were ion detector, all in series. The first quadrupole is set to scan
tested. First attempts were made using the method described for the molecular ions of the compounds of interest. Due to
by Tyczkowska et al.11 In short 0.5 g of milk was diluted the introduction of a collision gas into quadrupole 2 the ions
with an equal volume of a solution consisting of acetonitrile/ passed through quadrupole 1 are fragmented and daughter
methanol/water (40:20:20, v/v/v). The sample was vortex- ions are generated. The third quadrupole is set to scan for
mixed and placed in a microseparation system (Centricon one daughter ion of each compound. This gives an
10) and centrifuged. The ultrafiltrate was injected into the indication of which compound is present. Then a second
LC/MS system. We encountered problems with the injection is performed using a MS method following the
centrifugation procedure because the liquid could not pass transitions of the [M H] ion of that particular compound
through the cutoff filter of the Centricon 10. Therefore we to all the daughter ions (MRM). In this way it will probably
tried a clean-up by solid phase extraction based on the be possible to fulfil the criteria for confirmatory methods
method described by MacIntosh.16 To 2 g of milk were laid down by EU Decision 93/256.17
added 8 mL of 0.1 M ammonium acetate solution. A tC18 Recovery experiments were performed at three different
extraction column was conditioned with methanol and levels (MRL, 0.5 MRL, 2 MRL) in raw milk. The results are
ammonium acetate solution. The milk sample was applied presented in Table 2 and recoveries ranged from 63 to 87%
onto the column and passed slowly through the column. The at the MRL level. For amoxicillin, recovery experiments
extraction tube was rinsed with ammonium acetate, which were only performed at a concentration of 20 mg/kg and a
was also added to the column. The cartridge was dried by
passing air through it and elution was performed with 5 mL
of methanol. Water (1 mL) was added to the methanol and Table 3. Detection and quantification limits in retail and farm milk
the methanol was blown off under nitrogen. 20 mL of the Compound Detection limit (mg/kg) Quantification limit (mg/kg)
water phase was injected into the LC/MS/MS instrument. (MRL in mg/kg) Retail Farm Retail Farm
Recoveries ranged from 30 to 90% and the results obtained Penicillin G (4) 1.5 3 4 7
for repeatability were not satisfactory. Heller and Ngoh9 Ampicillin (4) 3 4 9 10
reported that a sample preparation based on simple Amoxicillin (4) 15 25 30 50
extraction with acetonitrile was quick but resulted in Cephapirin (10) 5 15 12 30
insufficiently cleaned extracts and was not suitable for Cloxacillin (30) 1 3 3 6
routine analysis. Nevertheless we tried this approach by Dicloxacillin (30) 2 3 4 8
adding 4 mL of acetonitrile to 2 g of milk. After vortex- Oxacillin (30) 1 1.5 3 4
mixing, the sample was centrifuged and after filtration Nafcillin (30) 2 2.5 4 5
20 mL of the supernatant were injected into the LC/MS/MS Cefazolin (50) 10 12 20 25
Ceftiofur (100) 1 3 3 6
system. With the Quattro LCZ instrument no problems were Cefalexin (100) 5 10 12 25
encountered with contamination of the interface and source.

Copyright # 2000 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 14, 14041409 (2000)
1408 DETECTION OF PENICILLINS AND CEPHALOSPORINS IN MILK
Figure 3. Standard curves in blank matrix for penicillin, cefazolin and oxacillin.
mean recovery of 57% was obtained. For most of the
compounds recovery is relatively constant and not depen-
dent on the concentration in the sample.
The results obtained for the LODs and LOQs are shown
in Table 3. In retail milk the detection limits are below the
MRL-values, except for amoxicillin. In farm milk problems
could arise, also, for cephapirin. The performance of the
developed method is comparable or better, in terms of
sensitivity, with other MS methods. Tyczkowska and
Voyksner reported thermospray (TSP)-LC/MS LODs for
penicillin of 100 mg/kg,11 for cephapirin of 500 mg/kg18
and for amoxicillin of 100200 mg/kg.10 Straub and
Voyksner reported an ESI-LC/MS procedure for determina-
tion of five b-lactams with a detection limit of 100 mg/kg.3
Boison et al. reported a TSP-LC/MS procedure with the
capability of confirming penicillin at 5 mg/kg in milk.19
Meetschen and Petz developed a GC/MS method for
b-lactams with neutral side chains;20 this procedure required
a very lengthy extraction procedure and derivatization but
was capable of confirming the compounds at a concentra-
tion of 4 mg/kg. Blanchflower et al. developed a procedure
for five b-lactams in muscle, kidney and milk.21 This
procedure employed ESI-LC/MS (negative ion mode) to
achieve detection limits of 5 mg/kg in milk, but was only
applicable to the non-amphoteric compounds penicillin and
cloxacillin. Keever et al. reported an ESI-LC/MS method
that could detect ceftiofur at 10 mg/kg in milk.22 Heller and
Ngoh reported an ESI-LC/MS method for the detection of
ampicillin, amoxicillin, cloxacillin, ceftiofur and cephapirin
at 10 mg/kg and penicillin at 5 mg/kg.9 In summary, the
LODs obtained with the method developed in this study are
similar to or, for some compounds, better than those
reported in the literature.
Linearity was checked by injecting blank milk samples
spiked with increasing amounts of the different standards
ranging from 0 to 150 mg/kg. The correlation coefficients
(R2) for the calibration curves were between 0.996 and
0.999 for all compounds. As an example the matrix
calibration curves for penicillin, oxacillin and cefazolin
are shown in Fig. 3.
Rapid Commun. Mass Spectrom. 14, 14041409 (2000)
From Table 2 it can be concluded that the coefficients of
variation obtained for repeatability at the MRL level are in
agreement with the criterion set by the Horwitz equation for
all compounds; i.e. a coefficient of variation has to be lower
than 11.5% for a concentration of 100 mg/kg, or lower than
16% for a concentration of 10 mg/kg. Specificity was
checked by analysing blank milk samples. No signals
greater than the LODs were detected at the relevant
retention times.
The results of the validation study show that this method
can be used to confirm the presence of 11 penicillins and
cephalosporins in farm and retail milk. Analyses of spiked
milk samples indicate that the method generates quantitative
results. Future work will address issues such as the
conformance of the developed method with the criteria for
confirmatory methods laid down in EU Decision 93/256 and
further validation and robustness testing of the method.
CONCLUSIONS
The ESI () LC/MS/MS method described in this paper
provides a sensitive and reliable procedure for the
determination of penicillins and cephalosporins in milk.
The simple extraction procedure involves a very short
sample preparation time and a high sample throughput with
acceptable recoveries, making the method very suitable for
routine analysis.
REFERENCES
1. Dewdney JM, Maes L, Raynaud JP, Blanc F, Scheid JP, Jackson T,
Lens S, Verschueren C. Fd Chem. Toxic. 1991; 29: 477.
2. EEC Council Regulation No. 2377/90, Off. J. Eur. Commun. 1990;
No. L 244/1.
3. Straub RF, Voyksner RD. J. Chromatogr. 1993; 647: 167.
4. Parker CE, Perkins JR, Tomer KB, Shida Y, OHara K. J.
Chromatogr. Biomed. Appl. 1993; 616: 45.
5. Tyczkowska KL, Voyksner RD, Straub RF, Aronson ALL. J.
AOAC Int. 1994; 5: 1122.
6. Straub R, Linder M, Voyksner RD. Anal. Chem. 1994; 66: 3651.
7. Moats WA, Harik-Khan R. J. AOAC Int. 1995; 78: 49.
Copyright # 2000 John Wiley & Sons, Ltd.
 DETECTION OF PENICILLINS AND CEPHALOSPORINS IN MILK
8. Moats WA, Romanowski RD, Medina MJ. J. AOAC Int. 1998; 81:
1135.
9. Heller DN, Nogh MA. Rapid Commun. Mass Spectrom. 1998; 12:
2031.
10. Voyksner RD, Tyczkowska KL, Aronson AL. J. Chromatogr.
Biomed. Appl. 1991; 567: 389.
11. Tyczkowska KL, Voyksner RD, Aronson AL. J. Chromatogr.
Biomed. Appl. 1989; 490: 101.
12. Rabbolini S, Verardo E, Da Col M, Gioacchini AM, Traldi P.
Rapid Commun. Mass Spectrom. 1998; 12: 1820.
13. Heller DN, Ngoh MA. Rapid Commun Mass Spectrom. 1998; 12:
2031.
14. Scandola M, Tarzia G, Gaviraghi G, Chiarello D, Traldi P. Biomed.
Environ. Mass Spectrom. 1989; 18: 851.
Copyright # 2000 John Wiley & Sons, Ltd.
1409
15. Tenconi S, De Filippo L, Da Col M, Cioacchini AM, Traldi P. J.
Mass Spectrom. 1999; 34: 268.
16. MacIntosh AI. J.Assoc. Off. Anal. Chem. 1990; 73: 880.
17. EEC Commission Decision 93/256, Off. J. Eur. Commun. 1993;
N L 118/64.
18. Tyczkowska KL, Voyksner RD, Aronson AL. J. Vet. Pharmacol.
Therap. 1991; 14: 51.
19. Boison JOK, Keng LJ-Y, MacNeil JD. JAOAC Int. 1994; 77: 565.
20. Meetschen U, Petz M. J. Assoc. Off. Anal. Chem. 1990; 73: 373.
21. Blanchflower WJ, Hewitt SA, Kennedy DG. Analyst 1994; 119:
2595.
22. Keever J, Voyksner RD, Tyczkowska KL. J. Chromatogr. 1998;
794: 57.
Rapid Commun. Mass Spectrom. 14, 14041409 (2000)