Journal of Anaivfi(al Toxi(ologv, \hi.

19, No~ember/De( ember 1995

Anticoagulant Poisoningin Animals:A Simple New

High-Performance Thin-Layer Chromatographic
(HPTLC) Method for the SimultaneousDetermination of
Eight Anticoagulant Rodenticidesin Liver Samples
Philippe J. Berny, Thierry Buronfosse,and Guy Lorgue
D~partement des Sciences Pflarmaceutiques et Toxicologiques, Ecole Nationale V~t~rinaire de Lyon, BP 83, F-69280 Marcy
I'Etoile, France
cation, and quantitation of anticoagulant rodenticides from
liver and serum samples will be presented in this paper.

Abstract
The purpose of this study was to develop and evaluate a technique
for the analysis of anticoagulant rodenticides in serum and liver
samples using a new high-performance thin-layer chromatographic
apparatus. Detection limits were estimated at 0.2 pg/g in liver
extracts for eight different substances. Overall, this technique was
repeatable and reproducible. The percent recovery was greater
than 87% for each substance. Liver and serum samples of animals
known to be exposed to one anticoagulant and showing clinical
signs of poisoning were analyzed. Concentrations measured varied
between 0.2 and 3 pg/g (liver extracts). Only blood samples from
one dog could be analyzed. The concentration was 150 ng/mL the
first day after admission and 140 ng/mL the following day.
Analyses are technically easily and rapidly performed, and they are
inexpensive. Therefore, this technique could be a valuable
alternative to current high-performance liquid chromatographic
methods.

Introduction
Anticoagulant rodenticides are among the most widely used
compounds to control rodent populations. However, many of
them are quite toxic to other mammals, especially companion
animals such as dogs and to a lesser extent cats, as well as wild
animals such as rabbits and birds of prey. They can be regarded as one of the major cause of poisoning in domestic
and wild animals (1,2). Several techniques have been developed
to analyze animal tissues to confirm poisoning cases. Liver is
used because it is known to accumulate anticoagulants for
days or weeks (3), and it is easily sampled to confirm anticoagulant poisoning in dead animals. However, liver biopsies are
not common practice in clinically affected animals, and less invasive techniques based on the determination of serum or
plasma anticoagulant concentrations can be very useful. The
purpose of this paper is to describe a new technique based on
high-performance thin-layer chromatography (HPTLC) because of its potential advantages over high-performance liquid
chromatographic (HPLC) methods. Mainly, isolation, identifi576

Materials and Methods

This technique used an HPTLC apparatus (automatic TLC
sampler III and TLC scanner II; Camag, Switzerland). Extraction was based on previously published methods (4). For routine purposes, it is essential to have efficient and rapid extraction procedures. For this reason, the technique was adapted as
follows (with ultrapure HPLC-grade reagents, unless otherwise specified): 1.0 g liver (2.0 mL plasma) was weighed and extracted with 5.0 mL acetone, ground, and then centrifuged at
2000 rpm for 5 min. The residue was extracted again with 4.0
mL acetone and centrifuged similarly. Both supernatant fractions were combined, and 1.0 mL diethylether was added to
eliminate protein extracts. After centrifugation (at 2000 rpm
for 5 rain), 1.0 mL supernatant was evaporated at 40~ under
a nitrogen flux and resolubilized with 100 ld, methanol (concentration coefficient, 1.0). Spiked samples were prepared to
determine the recovery, as follows: 1.0 mL of a 2% solution of
anticoagulant was added to 1.0 g liver sample and extracted in
a similar manner.
Standard solutions (2 pg/mL) were prepared in methanol
using the following substances: brodifacoum, bromadiolone,
chlorophacinone, coumatetralyl, coumachlor, difethialone,
difenacoum, and warfarin. These compounds were selected
because they are the only anticoagulant rodenticides marketed
in France. Calibration curves were prepared after dilution
(from 0.1 to 2 1Jg/mL). Standards were obtained from Lipha
(Lyon, France).
Sampling was then performed automatically, using an
automatic TLC sampler III (ATSIII). Ten microliters of each
sample or standard was sprayed as a thin band (6 x 0.5 ram)
with pressurized nitrogen gas (6000 hPa) 1.0 cm from the borders. Samples were separated by a 3-5-mm space. Usually
18-20 samples can be sprayed on one 20- 10-cm plate.
The plates used were nano HPTLC RP1~ plates (20 x 10 cm)

Reprodu(tion (photo(opyingl of editorial (ontent of this journal is F,rohibited without publishq-'r",permis,,ion.

Journal of Analytical Toxicology, Vol. 19, November/December 1995

(Merck, Nogent sur Marne, France). A single batch number was

used throughout the experiment. The plates were glass-coated,
reversed-phase silica gel plates, and the silica particles were
2-10-1Jm large.
After spraying, the plates were eluted in methanol-phosphoric acid (9:1). Phosphoric acid (4.72M) was prepared with
85% orthophosphoric acid in distilled water. A small elution
chamber was saturated for 5-10 rain with the solvent, and
elution took approximately 40 min (8-cm development). After
elution, the plate was removed and allowed to stand alone to
dry for 5-10 min. No further development was necessary.
Reading was performed with a TLC3 scanner (Camag), which
was driven by Cats software (Camag). A pilot track (standard)
was read first at 286 nm (deuterium lamp). This wavelength

was selected in preliminary experiments because the eight

substances analyzed absorbed well enough at 286 nm to enable
detection and quantitation. Peaks were then integrated. An
ultraviolet (UV) spectrum (220-380 nm in our experiment)
may also be obtained for each detected peak on the plate. Each
substance was then characterized by its retardation factor (Rr)
and UV spectrum. For each substance, three replicates per
plate were analyzed, and five plates were examined to check the
reproducibility of the Rrvalues measured.
Standard UV spectra were recorded in a spectrum library for
future identification purposes. Spectra of standards and sampies (unknown or spiked) were then compared by means of the
Cats software.
Calibration curves were prepared using five points (0.5-2.5
IJg/mL). For each point, three measurements were made to improve repeatability
~.88of the procedure. Reproducibility was
checked with three plates on three different
days. For each point, data were averaged,
88and the calibration curve was determined by
means of a linear regression model (leastE
squares regression). Extraction (percent rec
6it.
covery) was determined on six spiked samples for each substance (theoretical
N
concentration, 2 IJg/g). Dilutions of a 100._g
IJg/mL standard solution in methanol were
used.
Several samples were submitted to the
2~
laboratory after known exposure to one of
the eight aforementioned substances, which
occurred after accidental ingestion of a
8
8.8
8~5
~.o
rodenticide bait. Liver samples of dogs,
R;
rabbits, a turtle dove, pigeons, and cats were
analyzed. Also, serum samples from one dog
Figure 1. Chromatogram of a blank fresh liver extract. Arrows indicate the potential position (retaradmitted to the Veterinary School of Lyon
dation factor [Rt]) of the eight anticoagulant compounds marketed in France.
were analyzed after suspected exposure to
coumatetralyl.

, 11 1

lOO7

Results
(i)

g
g

fi

60q

Specificity
{~

/l

40
. D

/l

.... tI
I
I
I

J
0

1-

!
'

0.0
0.1
0.2
Wavelength:286nm

0.3

0.4

'

0.5

'

0.6

--

0.7

'

0.8

0.9

Rf

Figure 2. Chromatogram of 2-pg/mL solutions of three standards: (1) difethialone, (2) bromadiolone,
and (3) warfarin.

The specificity of the method was determined with blank liver and serum samples.
It appears that fresh liver extracts have very
few interfering peaks that could be misleading (i.e., the specificity is close to 100%)
(Figure 1). An example of a chromatogram
of standards is given in Figure 2 for bromadiolone, difethialone, and warfarin. If
liver extracts are not in good condition,
many interferences can be recorded, and
specificity is decreased. It is still possible to
scan each recorded peak (220-380 nm) to
determine whether the substance is a rodenticide or an interference. As shown in
Table I, anticoagulant rodenticides have suf577

Journal of Analvtk al Toxicolo,e,v,Vol. 19, .X,o~.en/l)er'De(enfl)er1995

limit of quantitation was determined as 0.5 lJg/mL (mean noise

level plus 10 standard deviations) for spiked liver samples.
Data were shown to follow a linear regression model between
0.5 and 2 lsg/mL (r 2, greater than 0.99), and the method appeared both repeatable and reproducible. The CVs were less
than 5% within plates and between plates for each of the eight
compounds tested at 2 1Jg/mL.

ficiently different Rrvalues to be well separated. These values

were found to be repeatable and reproducible from one plate to
another (coefficient of variation [CV], less than 5%). Each of
the eight substances tested had a specific UV spectrum. As an
example, the UV spectrum of chlorophacinone is shown in
Figure 3.
Extraction
The percent recovery varied between 85.7% (chlorophacinone) and 94.6% (coumatetralyl), as shown in Table II. The CV
was less than or equal to 5% for each compound.

Application to animal poisoning cases

Positive results were obtained for all 17 animals known to be
exposed to one anticoagulant compound (see Table III). When
the concentration could be measured, the value was given (in
micrograms per gram, wet weight basis). A plus sign means
that the substance was detected but not quantitated (between
0.2 and 0.5 lJg/mL). Liver concentrations were found between
0.2-0.5 and 14 lag/mL.
We also received serum samples from one dog known to be
exposed to coumatetralyl. Serum (2.0 mL) was extracted following the same procedure as described for liver extracts. The
only difference was that all supernatant fractions were mixed
and concentrated (20 times) to 0.1 mL. The plasma concentration went from 150 to 140 ng/mL from day 3 to day 4 after
the potential exposure. The dog was then treated for 3 weeks
and recovered (see Table III).
Based on these 18 animals, the sensitivity of the method
would be considered 100%. However, this value should be evaluated in field conditions to be more accurate.

Detection limit
The noise level was estimated on a blank liver sample, and
the mean noise level plus three standard deviations was used to
estimate the detection limit (5). In our test conditions, the detection limit was determined as 0.2 IJg/mL.

Quantitation--linearity, repeatability, and reproducibility

All of the anticoagulants tested gave similar results. The
Table I. Rt-Values for Eight Anticoagulant Compounds
Substance
Brodifacoum
Bromadiolone
Chloropha( inone
Coumatetraly[
Coumachlor
Difethialone
Difenacounl
Warfarhl

Mean Rp

CVt (%)

0.210
0.482
0.2/2
0.600

3.3
2.7
3.6
2.
2.1
2.8
2.9
2.6

0.632
0.158
0./04
0.717

Discussion

,Mt'an R~ on Ii\1, llhlle<< Ihree ri'lllk ,it(", I)('r san/pit' p(,i plat(,
( <>('lii( iiqll fll \ ,ilialion ll<l<>(,d on tn, l' I)lal( '<,.

100 -]
I
I
1

,!
/

80 ]

..I

'\

60 J

1I

40 j

/
\

,/

i1

J /

,~

I !

o L_/
22O

27O

320
VVa,.elengthInto,

Figure 3. Ultraviolet spectrum of chlorophacinone.

578

\
\

The use of HPTLC methodology has several advantages over

the usual HPLC techniques. The major advantages are its simplicity and rapidity. With one plate, 20 samples can be analyzed
within 1 h after the extraction has been
completed. It should be mentioned also that
the same apparatus can be used for many
other analyses (e.g., organophosphorus, carbamate insecticides, pyrethroid insecticides,
and heavy metals) as is routinely performed
in our laboratory, which is an invaluable
advantage over HPLC techniques. The use
of disposable, ready-to-use nano HPTLC
plates does not convey the problems pertaining to HPLC columns. Because of the
high quality of these plates compared with
standard TLC plates, this technique should
not be regarded as mere "efficient TLC" but
rather as an alternative to HPLC technology. Using a UV detector, it is possible
not only to detect a substance and quantitate it but also to identify this substance
'
with a high probability by means of a solid370
phase UV spectrum. This substantially increases the specificity of the method compared with standard TLC and allows for less
purification and separation phases. This

Journal of Analytical Toxicology, Vol. 19, November/December 1995

technique appears comparable with HPLC techniques using

diode-array detectors. Despite a high investment cost, HPTLC
is less expensive than HPLC on a routine basis (price/analysis). Many published techniques reported the use of silica or
C18 cartridges to purify the liver extracts (1,6-12).
Overall, this technique proved to be reproducible within
and between plates. The detection limit was comparable with
HPLC techniques (6,8,12) and gas-liquid chromatographic
and mass spectrometric methods (13,14). Using only acetone
Table II. Percen! Recovery from Liver Samples for Eight
Anticoagulants
Substance

% Recovery*

Brodifacoum
Bromadiolone
Chlorophacinone
Coumatetralyl
Coumachlor
Difethialone
Difenacoum
Warfarin

88.5
90.2
85.7
94.6
92.1
90.8
93.2
87.5

* The coefficient of variation was lessthan 5% on three replicates.

Table Ill. Liver Concentrations of Chlorophacinone,

Bromadiolone, and Brodifacoum and Serum
Concentrations of Coumatetralyl in Poisoned Dogs and
Wild Animals
Concentration
Species

Sample

Tissue

(pg/g)

Dog

Chlorophacinone
Chlorophacinone

Liver

1.7
2.1

Chlorophacinone
Bromadiolone
Brodifacoum

2.9
0.9
1.2

Rabbit (wild)

Chlorophacinone
CHorophacinone
Chlorophacinone
Chlorophacinone
Chlorophacinone
Chlorophacinone

Liver

2.0
4.4
5.1
14.3
2.7
3.0

Hare (wild)

Chlorophacinone
Chlorophacinone
Chlorophacinone
Chlorophacinone

Liver

3.2
+*
2.3

Pigeon

Chlorophacinone

Liver

1.7

Turtle d o v e

Chlorophacinone

Liver

1.8

Dog

Coumatetralyl (day 3)
Coumatetralyl (day 4)

Serum

0.150
0.140

* + = The compound was detected but not quantitated.

and diethyl ether, we obtained extracts pure enough to detect

0.2 ~g/mL anticoagulant rodenticide in 1 g of liver extract
without major interferences. It appears possible to lower the
detection limit by increasing the amount of acetone extract,
which is then evaporated and taken by methanol. Preliminary
experiments showed that using 2 g instead of 1 g, thereby doubling the concentration factor, would undoubtedly lower the
detection and quantitation limits. However, experimental and
field data in poisoned animals proved that, in rats, dogs, and
musk rats at least, liver concentrations of both indanedione and
coumarinic anticoagulants were greater than the 0.2-1Jg/mL
limit calculated in this paper (9). When there is clinical evidence of anticoagulant poisoning, published data on liver concentrations showed that they were usually much higher
(greater than 2 ~lg/mL) (3,9). Our data on several species of
mammals and birds were very similar and confirmed that liver
concentrations were usually well above our detection and
quantitation limits in clinical cases of anticoagulant poisoning.
These results also showed that there was not much variation
between species as far as liver concentrations were concerned.
Therefore, it does not seem necessary, for diagnostic purposes,
to increase the sensitivity of this technique.
Serum anticoagulant concentrations have been measured
and shown to decrease rapidly (10,15). The concentration of
brodifacoum was greater than 1 IJg/mL during the three days
after ingestion in dogs given 1.1 mg/kg per day for three days
(11). However, after the peak plasma concentration had occurred, concentrations remained approximately 10 ng/mL for
the following 20 days. In a human poisoning case, the initial
plasma concentration of brodifacoum was 731 ng/mL; 10 days
later, it was 240 ng/mL (15). These concentrations were associated with clinical signs of rodenticide poisoning (mainly
bleeding and bruising). It was also shown that serum concentrations of brodifacoum less than 10 ng/mL were associated
with normal clotting times in humans. We could only investigate one clinical case and found concentrations of coumatetralyl of 150 ng/mL, which were associated with clinical signs
of anticoagulant poisoning. Testing serum samples, in conjunction with clotting times, appears to be of great interest to
plan treatment strategies in clinical cases, as stated by
Hollinger and Pastor (15). In our laboratory, we mostly receive samples from dead animals because we are not part of an
emergency unit. Our limited data indicated that serum concentrations could be analyzed and could yield valuable information. Treatment, for instance, could be adapted based on the
compound involved and its blood concentration.

Conclusion
In many circumstances, it is necessary to quantitate anticoagulant residues in liver or serum samples. In this paper, we investigated the use of HPTLC on the livers of presumably poisoned animals. Our data on liver, as well as published literature
on liver analysis, showed that this technique has detection
limits and sensitivity comparable with most HPLC methods.
Specificity was increased using a IN spectrum between 200 and
400 nm, as compared with current fluorescence and UV de579

Journal of Analyti(al Toxi(ology,Vol. 19, November/December1995

tectors. Therefore, this technique could be used as an alternative to HPLC for the determination of liver anticoagulant concentrations. The preliminary data on serum extracts indicated
that this technique was quite applicable.

Acknowledgments
The authors wish to thank Mrs. J. Sinardet and D. Vey, as
well as Mr. A. Noel. This study would not have been completed
without their technical assistance and help throughout the
experiment.

References
1. G. Lorgue. Recent data concerning pesticides in veterinary clinical toxicology. Vet. Hum. Toxicol. 29 (suppl 2): 16 17 (1987).
2. G. Lorgue and A. Rivi@re.Les intoxications des animaux sauvages.
In Faune Sauvage d'Europe. R. Rosset, Ed., ITSV, Paris, France,
1987, pp 287-97.
3. M.F. Morin. Etude de I'impact sur le milieu naturel de la bromadiolone, rodonticide anticoagulant: @volution en milieu aqueux
et bioaccumulation par des organismes terrestres et aquatiques.
Ph.D. Thesis, Universit6 de Poitiers, France, 1988.
4. M.F. Morin, N. Merlet, M. Dore, and J.C. Lechevin. Analyse de
r~sidus de bromadiolone (rodonticide anticoagulant) dans ]'eau et
les tissus animaux. Techniques d'extraction et de purification.
Analysis 17:526-31 (1989).
5. J. CaporaI-Gautier, J.M. Nivet, P. Algranti, M. Guilloteau, M.
Histe, M. Lallier, J.J. N'Guyen-Huu, and R. Russotto. Guide de
validation analytique. Rapport d'une commission SFSTP I.

580

M6thodologie. S. lIP. Pharma Pratiques 2:205-26 (1992).

6. T. Chalermchaikit, L.J. Felice, and M.J. Murphy. Simultaneous
determination of eight anticoagulant rodenticides in blood serum
and liver. I. Anal. Toxicol. 17:56-61 (1993).
7. L.J. Felice and M.J. Murphy. The determination of the anticoagulant rodenticide brodffacoum in blood serum by liquid chromatography with fluorescence detection. ]. Anat. Toxicol. 13:
229-131 (1989).
8. M.J. Murphy, A.C. Ray, and E.M. Bailey. A high performance
liquid chromatography method for the detection of brodifacoum
in serum. Vet. Hum. Toxicol. 31:228-31 (1989).
9. A.C. Ray, M.J. Murphy, M.D. Duvall, and J.C. Reagor. Determination of brodifacoum and bromadiolone residues in rodent and
canine liver. Am. J. Vet. Res. 50:546-50 (1989).
10. H.J. Boermans, 1. Johnstone, W.D. Black, and M.J. Murphy. Clinical signs, laboratory changes and toxicokinetics of brodifacoum
in the horse. Can. J. Vet. Res. 55:21-27 (1991).
11. B.J. Woody, M.J. Murphy, A.C. Green, and R.A. Green. CoaguIopathic effects and therapy of brodifacoum toxicosis in dogs. J.
Vet. Int. Med. 6:23-28 (1992).
12. W. Langseth and U. Nymoen. Determination of coumarin anticoagulant rodenticides residues in animal liver by high performance liquid chromatography. Fresenius Z. Anal. Chem. 339:
249-52 (1991).
13. R.W. Bullard, G. Holguin, and J.E. Peterson. Determination of
chlorophacinone and diphenadione residues in biological materials. J. A[~ric. Food Chem. 23:72-74 (1975).
14. W.E. Braselton, R.G. Neiger, and R.P. Poppenga. Confirmation of
indanedione rodenticide toxicoses by mass spectrometry/mass
spectrometry. J. Vet. Diagn. Invest. 4:441~.6 (1992).
15. B.R. Hollinger and ].P. Pastor. Case management and plasma
half-life in a case of brodifacoum poisoning. Arch. Intern. Med.
153:1925-28 (19913).
Manuscript received January 5, 1995;
revision received March 113,1995.