Improvement of warfarin biopharmaceutics by conjugation with poly(ethylene glycol)

Article abstract of DOI:10.1016/j.ejps.2004.09.001

One of the most used and useful polymers, poly(ethylene glycol) (PEG) was used as a carrier for warfarin. The drug-polymer conjugate was freely water soluble at room temperature. The hydrolytic stability of the PEG-warfarin was investigated at physiological pH and confirmed the stability of the conjugate. In vivo release studies demonstrated a good release of parent drug, without the initial high plasma level of warfarin.

Full text of DOI:10.1016/j.ejps.2004.09.001

European Journal of Pharmaceutical Sciences 23 (2004) 379–384
Improvement of warfarin biopharmaceutics by conjugation with
poly(ethylene glycol)
Marina Zacchigna∗, Gabriella Di Luca, Francesca Cateni, Venerando Maurich
Dipartimento di Scienze Farmaceutiche, Piazzale Europa 1, 34127 Trieste, Italy
Received 4 August 2004; accepted 6 September 2004
Available online 28 October 2004
Abstract
One of the most used and useful polymers, poly(ethylene glycol) (PEG) was used as a carrier for warfarin. The drug–polymer conjugate was
freely water soluble at room temperature. The hydrolytic stability of the PEG–warfarin was investigated at physiological pH and confirmed
the stability of the conjugate. In vivo release studies demonstrated a good release of parent drug, without the initial high plasma level of
warfarin.
© 2004 Elsevier B.V. All rights reserved.
Keywords: Macromolecular prodrugs; Warfarin; Poly(ethylene glycol); Pharmacokinetic
1. Introduction
The FDA has approved this polymer for human intra-
venous, oral and dermal applications. In the case of PEG con-
jugates it is clear that the solubility of the prodrug will almost
always exceed that of the original drug, usually overcom-
ing any existing aqueous insolubility and thus increasing the
possibility of more effective drug delivery. The rate of drug
release will be dependent on the nature of the polymer–drug
linkage (Greenwald et al., 2003).
Warfarin (Fig. 1) is the most widely used coumarin an-
ticoagulant and is often considered as a drug of choice. If
warfarin is employed for a long term anticoagulant therapy,
it takes about 3 days for the prothrombin time to return to nor-
mal, after drug discontinuation. Warfarin is equally effective
either orally or intravenously, but is usually given by mouth.
Theoretically sudden discontinuation of warfarin may result
in rebound hypercoagulability with risk of thrombosis and
therefore some clinicians tail off long-term treatment over
several weeks (AA.VV., 1999).
The conjugation of simple drugs to synthetic polymers is
of great interest for pharmacological applications. Prodrug
design comprises an area of drug research that is concerned
with the optimization of drug delivery. A prodrug is a bi-
ologically inactive derivative of a parent drug molecule that
usually requires an enzymatic transformation within the body
in order to release the active drug and has improved delivery
properties over the parent molecule. Too rapid a breakdown
of the prodrug can lead to spiking of the parent drug and pos-
sible toxicity, while too slow a release rate can compromise
the drug’s efficacy (Greenwald et al., 2003).
Poly(ethylene glycol) or PEG, is one of the most used and
useful polymer, since it endows many advantageous features
both from a physico-chemical as well as from a biological
point of view. Indeed, PEG is amphiphilic and dissolves in
organic solvent as well as in water; it is non toxic and elim-
inated by a combination of renal and hepatic pathways thus
making it ideal to employ in pharmaceutical applications.
Warfarin is practically insoluble in water and it is ad-
ministered as warfarin sodium or warfarin sodium clathrate
(Babhair et al., 1985). Warfarin is a drug with an elimina-
tion half-life of 37 h, which is sufficiently sustained in the
body with conventional doses, so controlled release is gen-
erally not necessary. Controlled release of drugs that have
∗
Corresponding author. Tel.: +39 0405583104; fax: +39 04052572.
E-mail address: zacchign@units.it (M. Zacchigna).
0928-0987/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.ejps.2004.09.001

380
M. Zacchigna et al. / European Journal of Pharmaceutical Sciences 23 (2004) 379–384
2.3. Analytical methods
Infrared spectra were recorded using a Perkin-Elmer 1720
IR Fourier transform spectrophotometer. High-performance
liquid chromatographic (HPLC) analyses were carried out
with a Perkin-Elmer series 4-high-performance liquid chro-
matography, equipped with a Rheodyne Model 7125 injector
with a 20 ␮l loop and connected to a Perkin-Elmer Model
LC 75 variable-wavelength UV detector. A LiChrosorb RP18
(Perkin-Elmer) 250 mm × 4.6 mm i.d. column packed with
5 ␮m particle size was used.
Fig. 1. Warfarin.
The purity of the synthesized compounds was checked
by TLC on precoated silica gel aluminium sheets (Merck 60
long biological half-life is difficult to rationalize. The most
plausible reason is that by reducing the absorption rate it is
possible to prevent the occurrence of high peak drug lev-
els shortly after dosing and, therefore, the risk of adverse
effects. The current active interest in controlled drug re-
lease, together with improved technology, has given rise
to a rush of new formulations (Welling and Dobrinska,
1987).
As a part of our search for prodrug design (Zacchigna
et al., 2002, 2003), we report herein the design, synthesis
and evaluation of a macromolecular prodrug of warfarin, a
well known anticoagulant drug. The attachment of warfarin
to PEG, as a macromolecular carrier, was performed by
a covalent bond through succinic acid as spacer group.
Furthermore, in vitro and in vivo studies were carried
out in order to evaluate the behaviour of the obtained
prodrug. While several studies have been reported on
pharmacokinetics (Smith et al., 2000a,b; Stangier et al.,
2000) and formulation (Ammar et al., 1997) of warfarin
this is the first report on pegylation of this anticoagulant
drug.
F254). Spots were visualized under 254 nm illumination.
Potentiometric titrations were carried out using a TTT80
pH stat equipped with an ABU80 autoburette, a titrigraph
module REA 160 and a REA 070 pH stat unit, radiometer.
Elemental analyses were carried out on a Carlo Erba model
1016 analyser. 1H NMR and 13C NMR spectra (CDCl3) were
recorded on a Varian Gemini-200 MHz spectrometer.
13C NMR: 100.4 MHz. Chemical shifts are given in ppm
from TMS as internal standard. The following abbreviations
are used: s, singlet; br s, broad singlet; d, doublet; t, triplet;
m, multiplet; respectively.
2.4. Preparation of PEG–COOH
PEG2000 (10.5 g, 0.01 mol of OH groups) was dissolved
in toluene (100 ml) and dried by distilling off most of the
toluene. Succinic anhydride (4.2 g, 0.04 mol) was added and
the mixture was stirred for 5 h in an oil bath at 150 ◦C. The
mixture was cooled, taken up in CH2Cl2 and the polymer
precipitated by ether. The product was recrystallized twice
from CH2Cl2/ether. Yield 90%.
TLC: BuOH/AcOH/H2O, 4:1:1 v/v.
Analysis: calcd for C74H182: C, 40.36; H, 8.27, MW 2200.
Found: C, 40.52; H, 8.60.
2. Materials and methods
2.1. Chemicals
IR (nujol): ν = 1735 cm−1 (C O), 1100 (C O).
1H NMR (CDCl3): δ = 2.60 ppm (br s; 8H, CH2COO ),
3.60 (br s; CH2 O CH2), 4.20 (t; 4H, CH2OCO).
13C NMR (CDCl3): δ = 28.2–30.0 ppm (CH2COO), 64.1
(CH2OCO), 69.0–70.6 (CH2 O CH2), 172.0 (COO), 174.2
(COO).
Warfarin (3-(␣-acetonylbenzyl)-4 hydroxycoumarin),
warfarin sodium (3-(␣-acetonylbenzyl)-4 hydroxycoumarin
sodium salt), chloro warfarin (( )-3-(␣-acetonyl-p-chloro-
benzyl)-4 hydroxycoumarin), triethylamine (TEA) were
obtained from Sigma Chemicals Co. (St. Louis, MO, USA).
4-Dimethylaminopyridine (DMAP), poly(ethylene gly-
col) of MW = 2000 (n = 43), succinic anhydride, dicyclo-
hexylcarbodiimide (DCC) were purchased from Fluka.
All the other chemicals and solvents used in this work
were of the highest quality commercially available. Organic
2.5. Preparation of PEG–warfarin
Warfarin (6.8 mmol; 2.1 g), DCC (9 mmol; 1.86 g) were
added in small portions to a solution of PEG2000-(COOH)2
(4.5 mmol; 5 g) and DMAP (2 mmol; 0.44 g) in CH2Cl2
(50 ml); the mixture was stirred for 5 h at room tempera-
ture. The precipitate of dicyclohexyl urea (DCU) was filtered
off and the filtrate evaporated to dryness. The residue was
extracted with CH2Cl2, filtered from DCU residue and the
product precipitated by ether. The product was recrystallized
twice from ethanol. Yield 85%.
˚
solvents were dried over molecular sieves (3 A).
2.2. Laboratory animals
Male rabbits weighing 5 kg free from any signs of ab-
normality were used in this study (Charles, River, Calco,
Italy).

M. Zacchigna et al. / European Journal of Pharmaceutical Sciences 23 (2004) 379–384
381
Analysis. Calcd for C132; H210: C, 56.97; H, 7.55, MW
3. Results
2780; found: C, 56.82; H, 7.80.
1H NMR (CDCl3): δ = 2.06 ppm (s; 6H, CH3CO), 2.61
(br s; 8H, CH2COO), 3.46–3.60 (br s; CH2COCH3,
CH2 O CH2), 4.0 (t; 2H, CHPh), 4.20 (t; 4H, CH2OCO),
7.2–7.9 (m; 18H, arom-H).
3.1. Preparation of PEG–warfarin
A PEG–warfarin conjugate was synthesized in order to
study the role of PEG in drug delivery. An ester bond
links PEG to warfarin. Esters with PEG as an electron-
withdrawing substituent (alkoxy) in the ␣-position proved
to be especially effective linking groups in the design of
prodrugs since they aid in the enzymatic hydrolysis of the
ester carbonyl bond, and thus are able to release alcohols
in a continuous and effective manner (Greenwald et al.,
2003).
13C NMR (CDCl3): δ = 216.5 ppm (C-13), 176.9 (C-2),
171.9 (COO), 170.1 (COO), 160.7 (C-4), 152.9 (C-10), 142.4
(C-3), 131.9 (C-15), 128.8 (C-16, C-20), 127.0 (C-17, C-19,
C-7), 126.78 (C-18), 123.9 (C-6), 122.9 (C-5), 116.5 (C-8),
104.9 (C-9), 69.0–70.6 (CH2 O CH2), 64.0 (CH2 OCO),
41.7 (C-12), 35.15 (C-11), 28.9–29.8 (CH2COO), 24.4 (C-
14).
For the synthesis of warfarin prodrug, PEG2000 was used.
The hydroxylic groups of polymer were activated to car-
boxylic groups by succinic anhydride (90%) (Zalipsky et al.,
1983). TLC showed only one spot and no succinic anhydride.
The product titrated with 0.01 N NaOH yielded 100% of free
carboxylic groups.
2.6. In vitro drug release studies
An HCl, NaCl and glycine buffer (pH 1.2, 0.2 M), a phos-
phate/citrate buffer (pH 5.5, 0.2 M), and a phosphate buffer
(pH 7.4, 0.1 M) were used. To investigate the hydrolytic sta-
bility of PEG–warfarin. 20 mg/ml was dissolved in three
buffers (pH 1.2, pH 5.5 or pH 7.4), containing an appro-
priate quantity of internal standard at 37 0.1 ◦C. Samples
were taken at suitable intervals and the release of warfarin
was measured with HPLC.
The attachment of warfarin to the carboxylated polymer
was performed by means of DCC and DMAP with a good
yield (85%) (Fig. 2). The absence of free drug was verified
by HPLC analysis.
1H NMR and 13C NMR analysis were useful in providing
informations for the structural elucidation of the synthesized
drug-polymer conjugate.
2.7. In vivo drug release studies
The 1H NMR spectrum of the adduct indicated the
presence of methylene protons ascribed to PEG moiety
(δ = 3.46–3.60 ppm), whilethechemicalshiftsofthearomatic
protons of warfarin were observed at δ = 7.20–7.90 ppm. The
signals at δ = 171.9 ppm and 170.1 ppm in the 13C NMR spec-
trum of the conjugate confirmed the ester bond between the
drug and the activated polymer.
Warfarin sodium (0.54 mg/kg body weight) and
PEG–warfarin (2.3 mg/kg body weight) were solubilized in
physiological solution and administered orally, intravenously
or intramuscularly at three rabbits.
Blood (1.5 ml) was withdrawn immediately prior to drug
administration. Further 1.5 ml blood samples were obtained
at suitable intervals of time after drug administration. The
plasma was separated by centrifugation. Warfarin concen-
trations in plasma were measured with HPLC. The sam-
ple was prepared following a method already known in
literature (Bjornsson et al., 1977). Briefly: to 500 ␮l of
plasma in a capped test tube were added 500 ␮l of sul-
phuric acid 0.5 N containing 100 ␮l of p-chloro warfarin
(1 mg/ml) as internal standard and 5 ml of ether. After vor-
texing for 60 s, the samples were centrifuged for 5 min at
10,000 rpm. Four milliliter of the organic layer were trans-
ferred into a glass tube, and evaporated to dryness under a
nitrogen stream. The residue was reconstituted with 50 ␮l
of the methanol, and 20 ␮l of the solution were injected in
HPLC.
The percentage of linked drug, evaluated by alkaline hy-
drolysis followed by HPLC analysis, was estimated equal to
100% (w/w).
3.2. In vitro release studies
In order to gain some preliminary information about the
potential use of PEG–warfarin as a drug delivery system the
conjugate was subjected to hydrolysis at 37 0.1 ◦C in buffer
solutions at physiological pH, simulated gastric juice, pH 1.2,
endosomial compartments, pH 5.5, and extracellular fluids,
pH 7.4.
Samples were regularly taken out of the incubation mix-
ture and the percentage of released warfarin was quantified
with the HPLC. Each experiment was carried out in tripli-
cate and expressed as the mean value S.D. The results are
depicted in Fig. 3.
2.8. HPLC analysis
The amount of warfarin released from the polymeric con-
jugate at pH 7.4 in phosphate buffer solution was 2.5% over
24 h. In buffer solutions pH 5.5, and 1.2 less than 1.5% of
warfarin was released over 24 and 6 h, respectively. The con-
jugate therefore was hydrolytically stable in buffers under
physiological conditions.
Reverse phase HPLC C18 column was used. A mix-
ture of methanol and water (85:15 v/v %) pH 5 was
employed as mobile phase, and the detection wavelength
was 308 nm. p-Chloro warfarin was used as internal
standard.

382
M. Zacchigna et al. / European Journal of Pharmaceutical Sciences 23 (2004) 379–384
Fig. 2. Synthesis of PEG–warfarin.
3.3. In vivo release studies
The warfarin present in the plasma samples was quantified
by HPLC and the analysis of drug was validated. The limit of
quantification was approximately 0.1 ␮g/ml. The coefficient
of variation calculated for the six identical samples analyzed
did not exceed 5%. The precision of the assay method was
calculated by determining the relative standard deviations of
peak height ratios obtained from six replicate assays within a
concentration interval of 0.1–10 ␮g/ml. The relative standard
deviation ranged from 1 to 5% for intraday analysis and from
2 to 6% for interday analysis. The absolute recoveries of war-
farin and the internal standard in plasma were determined by
comparing the slopes of the processed human plasma stan-
dard curves and the standard curves prepared in methanol.
The recoveries of warfarin and the internal standard were
82 4% and 84 6%, respectively.
In order to evaluate the biological performance of the
adduct, a physiological solution of PEG warfarin, contain-
ing an amount of conjugate giving a dose of 0.5 mg/kg body
weight of warfarin, was administered in three separate ways:
orally, through a gastric tube and parenterally. Three male
rabbits fasted for 12 h prior to use, were used as subjects. The
group of animals was treated in the same way with a solu-
tion of warfarin sodium (0.54 mg/kg body weight) as control
drug. Each drug solution was prepared immediately prior to
use and the administration was carried out at a distance of 21
days in order to guarantee the complete elimination of war-
farin from the body. The blood samples were taken from the
right ear of the rabbits at regular time intervals. The plasma
was separated conventionally (Gasic et al., 1968) and stored
in a freezer pending assay.
Warfarin plasma concentration–time curves after oral and
parenteral administrations of a physiological solution of
Fig. 3. Release of warfarin from PEG–warfarin in buffer solutions, at 37 ◦C: (ꢀ) pH 1.2 0.2 M HCl, NaCl, and glycine, (ꢀ) pH 5.5 0.2 M phosphate/citrate,
and (ꢁ) pH 7.4 0.1 M phosphate.

M. Zacchigna et al. / European Journal of Pharmaceutical Sciences 23 (2004) 379–384
383
Fig. 4. Plasmatic concentration of warfarin after oral administration of (ꢀ) warfarin sodium and (ꢁ) PEG–warfarin. Each value is the mean S.D. for three
rabbits.
Fig. 5. Plasmatic concentration of warfarin after intramuscular administration of (ꢀ) warfarin sodium and (ꢁ) PEG–warfarin. Each value is the mean S.D.
for three rabbits.
macromolecular prodrug and of warfarin sodium are reported
in Figs. 4–6.
vides inadequate anticoagulation, too large a dose puts the
patients at risk of haemorrhage. This monitoring is com-
monly carried out by checking the prothrombin time. Mea-
surement should be carried out before treatment and then
daily or on alternate days in the early stages of treatment.
Once the dose has been established and the patient well sta-
bilized the measurement can be made at greater, but regular,
intervals. Dosage must be determined individually. A usual
initial dose of warfarin sodium is 10 mg daily by mouth for
2 days, but may be lower for patients at particular risk of
haemorrhage. Subsequent maintenance doses usually range
from 3 to 9 mg/die. These doses give steady state plasma
concentration of 0.72–2.9 ␮g/ml.
Warfarin sodium gives a plasma concentration of drug
that immediately increases after administration, but de-
creases very rapidly with time and remains fairly con-
stant. The drug plasma concentration after administration
of the polymeric prodrug is initially lower when compared
with the administration of free drug and decreases very
slowly with time, remaining practically constant for about
40 h.
4. Discussion
In this paper, the synthesis of a polymeric warfarin con-
jugate was described. Warfarin was covalently linked to the
macromolecular carrier, poly(ethylene glycol), via ester for-
mation.
Treatment with oral anticoagulants must be monitored to
ensure that the dose is providing the required effect on the
Vitamin-K-dependent clotting factors: too small a dose pro-

384
M. Zacchigna et al. / European Journal of Pharmaceutical Sciences 23 (2004) 379–384
Fig. 6. Plasmatic concentration of warfarin after i.v. administration of (ꢀ) warfarin sodium and (ꢁ) PEG–warfarin. Each value is the mean S.D. for three
rabbits.
The PEG–warfarin prodrug showed some interesting
peculiarities which make it attractive in the drug deliv-
ery for anticoagulant treatment. It was very soluble in
water and stable in physiological buffer, but, in vivo,
it was able to release drug in a constant and effective
manner. This is the first report of a PEG–warfarin pro-
drug.
Ammar, H.O., Ghorab, M., el-Nahhas, S.A., Makram, T.S., 1997. Im-
provement of biological performance of oral anticoagulant drugs. 1.
Warfarin. Pharmazie 52, 627–631.
Babhair, S.A., Tariq, M.A., Al-Bard, A.A., 1985. Warfarin. In: Florey,
K. (Ed.), Analytical Profiles of Drugs Substances, vol. 14. Academic
Press Inc., pp. 423–449.
Bjornsson, T.D., Blasche, T.F., Meffin, P.J., 1977. High-pressure liquid
chromatographic analysis of drugs in biological fluids I: Warfarin. J.
Pharm. Sci. 66, 142–145.
As the ratio of warfarin in the macromolecular conjugate
is almost 25%, 12–36 mg/die doses of the prodrug could give
the same plasma concentration at the steady state, without
the high drug concentration found immediately after the ad-
ministration of warfarin sodium. As the fluctuation of the
concentration at steady state is reduced and the therapeu-
tic concentration is more linear and constant, the risk of ad-
verse effects, like haemorrhage, can be reduced. In such cases
the checking of the prothrombin time could be less frequent
and consequently the compliance of the patient could be in-
creased.
Gasic, G., Gasic, T.B., Steward, C.C., 1968. Antimetastatic effects as-
sociated with platelet reduction. Proc. Natl. Acad. Sci. U.S.A. 61,
46–52.
Greenwald, R.B., Choe, Y.H., McGuire, J., Conover, C.D., 2003. Effective
drug delivery by PEGylated drug conjugates. Adv. Drug Deliv. Rev.
55, 217–250.
Smith, S.A., Kraft, S.L., Lewis, D.C., Freeman, L.C., 2000a. Plasma
pharmacokinetics of warfarin enantiomers in cats. J. Vet. Pharmacol.
Ther. 23, 329–337.
Smith, S.A., Kraft, S.L., Lewis, D.C., Melethil, S., FKreeman, L.C.,
2000b. Pharmacodynamics of warfarin in cats. J. Vet. Pharmacol. Ther.
23, 339–344.
Stangier, J., Su, C.A., Hendriks, M.G., van Lier, J.J., Sollie, F.A., Oost-
erhius, B., Jonkman, J.H., 2000. Steady-state pharmacodynamics of
warfarin in presence and absence of telmisartan in healthy male vol-
unteers. J. Clin. Pharmacol. 40, 1331–1337.
Welling, P.G., Dobrinska, M.R., 1987. In: Robinson, R.J., Lee, V.H.L.
(Eds.), Controlled drug delivery: Fundamentals and Applications, vol.
29. Marcel Dekker Inc., New York, pp. 253–291.
Zacchigna, M., Di Luca, G., Cateni, F., Zorzet, S., Maurich, V., 2003.
Improvement of physicochemical and biopharmaceutical properties
of theophylline by poly(ethylene glycol) conjugates. Il Farmaco. 58,
1307–1312.
Acknowledgements
The authors would like to thank M.U.R.S.T. for its support
of this work. Special recognition is extended to dott. Sonia
Zorzet for her technical support.
Zacchigna, M., Di Luca, G., Maurich, V., Boccu`, E., 2002. Synthe-
ses, chemical and enzymatic stability of new poly(ethylene glycol)-
acyclovir prodrugs. Il Farmaco. 57, 207–214.
References
Zalipsky, S., Gilon, C., Zilkha, A., 1983. Attachment of drugs to polyethy-
lene glycols. Eur. Polym. J. 19, 1177–1183.
AA.VV., 1999. In: Parfitt, K. (Ed.), Martindale. The Complete Drug Ref-
erence, 32nd ed. Pharmaceutical Press, London, pp. 964–971.