Carboplatin decomposition in aqueous solution with chloride ions monitored by X-ray absorption spectroscopy

Article abstract of DOI:10.1039/b006374f

Carboplatin aqueous solutions, with chloride ions added at different concentrations, were studied by X-ray absorption spectroscopy (XAS). The comparison of solid and solution spectra shows that carboplatin and cisplatin spectra are strongly different, and that the carboplatin ligands induce a specific structure of the spectrum, conserved in solution. Hence, it is possible to study by XAS the evolution of carboplatin in solution. This study shows that carboplatin is the major compound present in solution, even after 15 days, in neutral solutions with chloride concentration less than 9%, exposed to light or not. On the contrary, with high chloride concentrations (18%) or in acidic solutions (0.1 M HCl), the carboplatin is chlorolysed, the evolution of the solution composition can be followed by XAS and cisplatin formation is evidenced.

Full text of DOI:10.1039/b006374f

View Article Online / Journal Homepage / Table of Contents for this issue
Carboplatin decomposition in aqueous solution with chloride ions
monitored by X-ray absorption spectroscopy
Emmanuel Curis,*ab Karine Provost,c Ioannis Nicolis,b Diane Bouvet,c Simone Be
Sylvie Crauste–Manciet,de Francoise Brionfe and Denis Brossardde

nazeth,ab
Ó
a L URE, Ba
E-mail: curis=lure.u-psud.fr
b L aboratoire de Biomathematique, Faculte
avenue de lÏObservatoire, 75006 Paris, France
c Groupe de Physique des Milieux Denses, Faculte
XII, 61 avenue du General de Gaulle, 94040 Creteil, France
d Service de Pharmacie, CHI Poissy Saint-Germain-en-L aye, 20, rue Armagis,
8105 Saint-Germain-en-L aye, France
e L aboratoire de Pharmacie Galenique, Faculte
avenue de lÏObservatoire, 75006 Paris, France
f L aboratoire de Pharmacie Clinique, Faculte de Pharmacie, Universite
avenue de lÏObservatoire, 75006 Paris, France
ü
timent 209D, Centre Universitaire Paris Sud, 91405 Orsay cedex, France.

de Pharmacie, Universite Rene Descartes, 4,
  

de Sciences et T echnologies, Universite Paris




7


de Pharmacie, Universite


Rene Descartes, 4,


Rene Descartes, 4,

Received (in Montpellier, France) 2nd August 2000, Accepted 16th August 2000
First published as an Advance Article on the web 22nd November 2000
Carboplatin aqueous solutions, with chloride ions added at di†erent concentrations, were studied by X-ray
absorption spectroscopy (XAS). The comparison of solid and solution spectra shows that carboplatin and cisplatin
spectra are strongly di†erent, and that the carboplatin ligands induce a speciÐc structure of the spectrum, conserved
in solution. Hence, it is possible to study by XAS the evolution of carboplatin in solution. This study shows that
carboplatin is the major compound present in solution, even after 15 days, in neutral solutions with chloride
concentration less than 9%, exposed to light or not. On the contrary, with high chloride concentrations (18%) or in
acidic solutions (0.1 M HCl), the carboplatin is chlorolysed, the evolution of the solution composition can be
followed by XAS and cisplatin formation is evidenced.
Cisplatin (Fig. 1) Ðgures among the Ðrst antitumoral drugs
used. It o†ers a wide Ðeld of therapeutic applications and is
quite useful, but it presents a great nephrotoxicity. Because of
this, some new molecules have been synthesised. A particu-
larly used one is carboplatin (Fig. 1), which presents the same
Ðeld of e†ects but is much less toxic. However, the stability of
this compound in solutions and bags for injections is ques-
tioned. In particular, if the solution contains some chloride
anions, as may be the case during drug administration, chlo-
rolysis of the carboplatin into cisplatin could occur.
The question of the stability of the carboplatin has already
been approached, using techniques like HPLC or electro-
phoresis.1h6 These di†erent studies showed that carboplatin
slowly decomposes into a family of di†erent species; thus, the
injectable solutions must be prepared without chloride anions.
The nature of the di†erent degradation products is not well
deÐned, and seems to depend on the light exposure of the
solution.5 The detection methods used in these studies pre-
sented the drawback of being indirect methods of studying the
platinum environment, and in particular the HPLC column
itself may lead to some of the observed degradation products.
NMR spectroscopy gives direct information about the plati-
num co-ordination; it has been used to investigate similar
platinum complexes either in the solid state (for instance, by
13C experiments where the authors followed the Ñexibility of a
carboplatin derivative)7 or in solution by 195Pt experiments
where cisplatin and transplatin hydrolysis was analysed before
its interaction with DNA.8 This last NMR technique presents
the disadvantage of needing a special enrichment of the
sample in 195Pt and so prevents any work on the drug itself.
For the title study, we have applied X-ray absorption spec-
troscopy (XAS), which is a technique that probes speciÐcally
the local environment of the platinum ion and that may be
applied directly on the carboplatin drug solutions as used in
pharmaceutical applications. During the experiments here pre-
sented, and taking into account the sensitivity of this tech-
nique, the information obtained will concern platinum
complexes present in the acid solutions as major components.
We Ðrst studied the solid forms of carboplatin and cisplatin,
to quantify the di†erences between their EXAFS features and
compare them with their solutions. This preliminary work
showed that EXAFS spectra present, as expected, strong dif-
ferences between the two compounds. We then prepared
several aqueous solutions of carboplatin, under diverse pH
conditions and chloride concentrations, and we followed the
evolution of these solutions by XAS. Among these, the spectra
of the hydrochloric acid solutions showed a quick evolution
from the carboplatin spectrum towards the cisplatin one.
The aim of this work is to determine under which physical
and chemical conditions carboplatin can lead to cisplatin, and
Fig. 1 Formulas of cisplatin (left) and carboplatin (right).
DOI: 10.1039/b006374f
New J. Chem., 2000, 24, 1003È1008
1003
This journal is ( The Royal Society of Chemistry and the Centre National de la Recherche ScientiÐque 2000

View Article Online
to check the sensitivity of the XAS method for this kind of
system. During further studies, we will study these solutions
under the well-deÐned conditions of therapeutic drug conser-
vation and administration.
XAS measurements
All the spectra, except the cisplatin solution, were recorded at
LURE (Orsay, France), on the EXAFS 13 experimental
station (beamline D41, DCI storage ring, with a current of
about 250 mA), at the L platinum edge (11 564 eV). A Si(1 1 1)
III
Experimental
double-crystal monochromator was used; the energy was cali-
brated using the L edge of a gold foil. At this energy, and on
this storage ring, there is no need to eliminate harmonics. The
III
Sample preparation
cisplatin solution was studied on the EXAFS 2 experimental
station of LURE. On this station, it was not possible to use
the Si(1 1 1) monochromator, and we used a Si(3 1 1) crystal;
the calibration was done with a gold foil, and no harmonic
rejection was needed.
The solid samples were prepared as pellets. The pellets were
prepared so that the platinum concentration leads to an edge
jump of about unity at the platinum L edge, and completed
III
to 60 mg with cellulose. Cisplatin was purchased from Sigma
Chemicals; the carboplatin used was obtained from a phar-
maceutical preparation from BristolÈMyers Squibb
The cisplatin solution was studied by Ñuorescence mode,
because of the dilution of the sample. The detector was a 7-
elements Ge detector from Eurisys Mesures, placed at about
5 cm from the sample to avoid detector saturation (we
(
““paraplatineÏÏ, ref. 331 720.0).
We also prepared a saturated aqueous solution of cisplatin,
by adding solid cisplatin to water until no more dissolution
occurred, and Ðltering.
observed less than 5% of dead time). The L Ñuorescence line
was selected by electronic Ðlters (SCA). This cisplatin solution
III
The details of the di†erent carboplatin solutions used is pre-
sented in Tables 1 (neutral solutions) and 2 (acidic solutions).
Each solution was prepared from pharmaceutical ready-to-
inject paraplatine (10 mg mL~1 of carboplatin solution, which
gives nearly 2.7 ] 10~2 mol L~1 of platinum). For the XAS
measurements, 0.1 mL of each solution was placed in a TeÑon
cell, closed with Kapton. For samples stored in the dark, the
cell was also protected from light with an aluminium foil,
was also studied in transmission mode, under the same condi-
tions as the carboplatin solutions (see below), to check that
the contribution of this low solubility cisplatin solution can be
neglected during a transmission mode spectrum.
All the other samples were exclusively studied in the trans-
mission mode, the X-ray intensities being detected by two ion-
isation chambers Ðlled with air. Measurements where made
from 11 400 to 12 600 eV with a step of 3 eV and a counting
time of 1 s per point. The resulting recording time was about
20 min for each spectrum.
nearly transparent to X-rays around the platinum L edge.
III
Solutions S to S (aged solutions) were prepared 15 days in
1
5
advance, so that eventual degradation products would be in
sufficient concentration to be detected by XAS. The other
solutions were prepared just before the measurements.
For the reference samples (solids) and the aged solutions (S
1
to S ), for which no evolution was expected, two spectra were
5
acquired.
After 15 days, the two 18% NaCl solutions (S and S )
For the fresh solutions (S to S ) expected to evolve, the
spectra were recorded repeatedly during either at least half a
day or until we got a stable absorption spectrum.
5
d
5l
6
12
gave an important yellow precipitate. This precipitate was iso-
lated by centrifugation, dried for three days in a drying oven
and pressed into a pellet.
The acidic solutions quickly lead to a similar yellow precipi-
tate, within the experimental time.
XAS analysis
The XAS analysis has been performed using a classical model
with LASE (Logiciel dÏAnalyse des Spectres EXAFS), devel-
oped by one of us9 and o†ering new tools for error estimation.
When possible (solid samples, solutions showing no evolution
within the experiment time), averages where made at the very
beginning of the data treatment, on the experimental absorp-
tion coefficient. Error bars were computed simultaneously and
propagated all along the data analysis, including Fourier Ðl-
tering, according to the model developed by some of us.10
All the data treatment used a k2 weighting of the EXAFS
oscillations (k is the wave vector), to compensate for the
decrease of the oscillations as k increases. The conversion to
R-space (R is the inter-atomic distance) was achieved by using
a Fourier transformation (FT) with a KaiserÈBessel function
as the windowing function, between 1 and 14 AŽ ~1. All the FT
Ðgures presented in this paper are uncorrected for phase shifts
and so the R distances are just apparent path lengths, smaller
than the true values (even \2 AŽ ).
As expected, strong di†erences are observed between the
carboplatin and the cisplatin spectra. To explain these di†er-
ences, we used the known structures of cisplatin11 and
carboplatin12 to create a structural model. That model was
computed using the FEFF 7.02 software.13 We detail the new
experimental spectra Ðtting procedure we have developed in
LASE. Using the classical least-squares algorithm we mini-
mised the sum S deÐned by
Table 1 Composition of the di†erent neutral carboplatin solutions
studied. All were made by adding 4 mL of paraplatine to the indicated
mass of solid NaCl
Solution
NaCl/mg
Storage condition
S
0 (0%)
0 (0%)
Light
Dark
Light
Dark
Light
Dark
Light
Dark
Light
Dark
Direct usage
Direct usage
Direct usage
1
1
2
2
3
3
4
4
5
5
6
7
8
l
S
S
S
S
S
S
S
S
S
S
S
S
d
l
13.1 (0.45%)
13.1 (0.45%)
36.3 (0.9%)
36.3 (0.9%)
73.3 (1.8%)
73.3 (1.8%)
880 (18%)
880 (18%)
0 (0%)
d
l
d
l
d
l
d
36.3 (0.9%)
234 (5.8%; 1 M)
Table 2 Composition of the di†erent acidic carboplatin solutions
studied. All were made by adding the indicated volume of HCl to
1
mL of paraplatine
Initial
[HCl]/mol L~1
Final
[Cl~]/mol L~1
Solution
HCl/mL
S \ 1
;
N
G
s
(k ) [ ;
exp i
M N f (k@)
j j i
S
9
S
S
0.0254
0.0508
0.254
0.1
1.97
1.97
1.97
11.5
0.05
0.1
0.4
1
N
k@ R2
S
i/1
j/1
i j
1
1
1
0
1
2
C
R
j(k@)
D
H
2
]
exp [2p2k@2 [ 2
j
sin[2k@ R ] / (k@)]
j i
j i
i j
i
1004
New J. Chem., 2000, 24, 1003È1008

View Article Online
where N is the number of experimental points in the Ðltered
spectrum and M is the number of shells used in the model.
The electronic terms f (k@), / (k@) and j(k@) were computed by
j i
j i
i
FEFF, whereas the structural parameters were Ðxed (N ) or
j
were allowed to vary while Ðtting (p2 and R ). Because of the
uncertainties in the edge energy E determination, one also
j
j
0
introduces *E \ E
[ E
as a Ðt parameterÈwhich
0
0, exp 0, FEFF
leads to the use of k@ instead of k . The errors on the Ðtted
i
i
values were determined with the help of a MonteÈCarlo simu-
lation included in the LASE software. These errors only
account for the statistical uncertainties induced by the sta-
tistical Ñuctuations of the di†erent recordings. Hence, one can
obtain very small errors bars; the real uncertainties are prob-
ably somehow higher, because of systematic error e†ects, not
taken into account in this method. On this delicate question
of error estimation in EXAFS analysis our work is still in
progress.
Results and discussion
We Ðrst present the results comparing the EXAFS spectra of
the platinum complexes in the solid state and aqueous neutral
solutions excluding chloride ions.
Carboplatin complex
Carboplatin solid. The FT of the EXAFS spectrum of solid
carboplatin is reported in Fig. 2(a). We observe a narrow main
peak, due to the Ðrst coordination shell, and minor peaks at
greater distances. The theoretical model conÐrms that carbo-
platinÏs Ðrst peak FT comes from a single scattering shell, con-
taining four light atoms like oxygen or nitrogen [as the
electronic terms f (k) and /(k) are very similar for these two
kinds of atom, XAS cannot distinguish them] at an average
distance of 2.0279 ^ 0.0002 AŽ . This result is coherent with the
known structure of carboplatin, in which platinum is linked to
two nitrogen and two oxygen atoms located at 2.021 ^ 0.008
and 2.025 ^ 0.006 AŽ , respectively.12 The complete results are
given in Table 3. The very small value of the error bar results
from the small statistical uncertainty and probably underesti-
mates the real error because of systematic errors; the same
e†ect arises for the other results presented in this article.
To explain the other features of the EXAFS spectrum, we
started with a simple model of multiple scattering in the Ðrst
co-ordination shell of the platinum atom: a planar square
with two oxygen and two nitrogen atoms. The FEFF compu-
tations show that, among the diverse multiple scattering
paths, the most important is the Pt ] N ] Pt ] O ] Pt
Fig. 2 (a) Comparison of the experimental (solid line) and theoretical
(
]]]) FT (modulus and imaginary part) of the EXAFS spectrum of
solid carboplatin. The main features of the FT are reproduced by the
model. Some long distance contributionsÈprobably involving the
platinum atoms of neighbouring moleculesÈare not model. (b) Com-
parison of the FT (modulus) of the EXAFS spectra of carboplatin in
the solid form (solid line) and in aqueous solution (squares). The
spectra are nearly identical.
1
2
path, the nitrogen and oxygen atoms being diagonally
opposed. The second most important path is the
Pt ] N ] O ] Pt path. All other paths are much less
1
2
important and can be neglected. This result is coherent with
the well-known e†ect of atom alignment being related to the
increased contribution of multiple scattering paths.
role. The comparison of the theoretical spectrum and the
experimental spectrum is shown on Fig. 2(a).
With the help of the crystallographic co-ordinates of carbo-
platin,12 we built a quite complete model to reproduce the
main features of the full spectrum of the carboplatin in its
solid form. This modelÏs results are shown in Table 4; because
of the previous result, we only include the most signiÐcant
multiple scattering paths. The results show that the scattering
by the cyclobutanedicarboxylate atoms plays an important
One can note the high value for *E , close to [12 eV. This
0
is not surprising, since the edge white line is due to a 2p ] 5d
transition, the 5d level being around 8 eV below the ionisation
energy.14 Taking the edge energy E in the white line rising
0
necessarily leads to an incorrect E value. But, since this work
0
Table 4 Results of the Ðtting of the complete spectrum of carbo-
Table 3 Results of the Ðtting of the Ðrst peak of carboplatin. A single
platin. The Ðnal error is S \ 0.012; *E \ [12.3 ^ 0.2 eV
0
shell including two oxygen atoms and two nitrogen atoms is sufficient
to model the spectrum. The Ðnal error is S \ 1.03 ] 10~3 and *E \
Path
N (Ðxed) R/A
Ž
p2 ] 103/AŽ 2
0
[7.98 ^ 0.04 eV
Pt ] N/O ] Pt
4
2
1
1
2.028 ^ 0.002 2.00 ^ 0.14
Path
N (Ðxed)
R/A
Ž
p2 ] 103/A
Ž
2
Pt ] C ] Pt
2.94 ^ 0.06
3.16 ^ 0.14
20 ^ 60
80 ^ 18
2.5 ^ 1.4
1
Pt ] C ] Pt
2
Pt ] N/O ] Pt
4
2.0279 ^ 0.0002
1.91 ^ 0.02
Pt ] N ] Pt ] O ] Pt
4.067 ^ 0.006
1
2
New J. Chem., 2000, 24, 1003È1008
1005

View Article Online
compares the XAS spectra of two compounds and since one
can expect the same e†ect in the cisplatin spectrum, it is
important to have a precise deÐnition of the edge energy, and
the common deÐnition used in this work is more convenient.
of a speciÐed atom in solution, and is not limited to the Ðrst-
shell analysis.
Cisplatin complex
Cisplatin solid. The FT of the EXAFS spectrum of solid cis-
platin is reported in Fig. 4(a). On this FT we notice a much
broader main peak than in the case of carboplatin. This
strong di†erence was expected as this peak results from two
interfering contributions due to the nitrogen and chloride co-
ordination atoms. Two smaller peaks are present at greater
distances.
We thus considered a two-shell model to reproduce this
nearest cisplatin co-ordinationÈsee Table 5 for the complete
results. The Ðrst shell contains two nitrogen atoms at
Carboplatin aqueous neutral solution. The spectrum of the
fresh aqueous solution of carboplatin (solution S ) did not
6
show any signiÐcant change from the Ðrst spectrum to the last
one (with 9 h delay between them). Hence, we averaged all the
spectra and compared the average EXAFS oscillations and
Fourier transform to the solid carboplatin spectrum [Fig.
(b)]. This comparison shows that, under these conditions and
the stated time interval, carboplatin is stable in aqueous solu-
tion, at the detection level of XAS.
2
One can note that not only the Ðrst peak, but also the
weaker second peak, is conserved in solution. That means that
the complex is also well-deÐned in solution. This result is con-
Ðrmed by the comparison of the model obtained for the solid
and the spectrum obtained from the solution [Fig. 3(a) and (b)
respectively in k and R spaces]. This shows that X-ray absorp-
tion spectroscopy is a powerful tool to study the environment
2
.032 ^ 0.007 A
atoms at 2.314 ^ 0.004 A
structure of cisplatin,11 in which 2.00 ^ 0.04 A
distance and 2.33 ^ 0.01 A for the PtÈCl distance were found.
Ž
and the second shell contains two chloride
. This is coherent with the known
for the PtÈN
Ž
Ž
Ž
Fig. 4 (a) Comparison of the FT (modulus and imaginary part) of
the EXAFS spectrum of solid cisplatin (solid line) and the model
(
squares). We assigned all the peaks by studying the multiple scat-
tering paths and molecular packing. (b) Comparison of the cisplatin
EXAFS FT in the solid state (]]], in transmission mode) and
neutral aqueous solution (solid line, in Ñuorescence mode). One can
note, for the solution, in the Ðrst main peak, a decrease of the longer
distance contribution, which might reveal the onset of cisplatin hydro-
lysis.
Fig. 3 Carboplatin in aqueous solution (solid line) compared with
the theoretical model obtained for the solid (]]]). (a) k2s(k),
EXAFS normalised. (b) FT (modulus and imaginary part). The model
is still correct.
1006
New J. Chem., 2000, 24, 1003È1008

View Article Online
Table 5 Results of the Ðtting of the Ðrst peak of solid cisplatin. Two
shells are needed, one with the two nitrogen atoms and one with the
two chlorine atoms, to model the spectrum. *E was constrained to
0
be identical for the two shells and was [12.9 ^ 0.1 eV. The Ðnal
error is S \ 2.03 ] 10~3
Path
N (Ðxed)
R/A
Ž
p2 ] 103/AŽ 2
Pt ] N ] Pt
Pt ] Cl ] Pt
2
2
2.032 ^ 0.007
2.314 ^ 0.004
3.3 ^ 1.0
3.4 ^ 0.4
Using this crystallographic structure, we have demonstrated
that the peak observed on the FT [Fig. 4(a)] at a 3.6 A appar-
Ž
ent path length is mainly due to multiple scattering through
the platinum atom: Pt ] N ] Pt ] Cl ] Pt. The peak at
nearly 3 A
Ž apparent path length is assigned to a PtÈPt signal
related to the packing in the solid and we have checked that
this contribution disappears when considering
complex model.
a single
Cisplatin neutral aqueous solution
The recording of the solution spectrum has been tested in
transmission mode but the Pt L edge jump is then too small,
III
in relation with its low solubility: even for this saturated solu-
tion the height is near 0.04 (the recommended value to allow a
good extraction of the EXAFS spectra is between 0.4 and 1).
During our further study on carboplatin solution stability we
worked in transmission mode, and so an eventual contribu-
tion of cisplatin in solution, even in its hydrolysed form, will
not produce a signiÐcant contribution to the detected signal.
We have thus studied cisplatin solution in Ñuorescence mode
[Fig. 4(b)]. As usual the quality of the measurements is then
reduced [error bars increase in Fig. 4(b) relative to Fig. 4(a)]
and we just give a qualitative analysis.
When we compare solid and solution spectra, we note dif-
ferences in the FT spectra. The peak at 3 AŽ has disappeared.
This is not surprising as it is related to PtÈPt intermolecular
interactions. The main peak still contains two contributions
but their relative ratio has changed: the chloride contribu-
tions is less apparent than in the solid form. This might reveal
the onset of cisplatin hydrolysis but we have not analysed this
phenomenon more deeply: the investigation of this delicate
question is not the aim of this paper and has been extensively
studied.8,15h17 The purpose of this paper is to follow the car-
boplatin decomposition into cisplatin in chloride-containing
media, which is the subject of the remainder of this paper,
after these preliminary results.
Fig. 6 Comparison of the spectra of solid cisplatin (solid line, with
error bars) and solution 12 (supernatent) (squares). (a) k2s(k), normal-
ised. (b) FT modulus. The spectra are identical; the degradation
product of carboplatin is cisplatin.
Fig. 5 Evolution of the FT modulus of the EXAFS spectrum of a
carboplatin solution in 1 M HCl. Squares: solid carboplatin, dia-
monds: solid cisplatin. Solid line: t \ 2 min; crosses: t \ 40 min;
dashed line: t \ 120 min.
Fig. 7 Evolution of the cisplatin (squares) and carboplatin (circles)
molar fraction in 0.4 M HCl solution, obtained by linear decomposi-
tion of the EXAFS spectra (see text). The conversion of carboplatin
into cisplatin is evident.
New J. Chem., 2000, 24, 1003È1008
1007

View Article Online
methods to determine a and b from the experimental spectra;
for instance, a least-square minimisation of
N
S@ \ ; [s (k ) [ as
(k ) [ bs (k )]2
exp i
CisPt i CbPt i
i/1
allowing a and b to vary, using the solid state or the aqueous
solution spectra of carboplatin and cisplatin as references, and
we e†ectively observed that a increases with time. Fig. 7 shows
the result of such an analysis for the 0.4 M solution, when
working with a k2 weighting: the conversion of carboplatin
into cisplatin is evident. The fact that a ] b is around unity,
without constraints, conÐrms the validity of the decomposi-
tion hypothesis. However, the obtained values cannot be used
as quantitative results, because of the heterogeneity we have
mentioned previously, but they give a main tendency of the
carboplatin into cisplatin conversion and can be used to
compare qualitatively the evolution of di†erent solutions.
Fig. 8 compares the values obtained for the 0.4 M and the
Fig. 8 Evolution of the carboplatin molar fraction in 0.4 M (circles)
and 0.05 M (squares) HCl solutions, obtained by linear decomposition
of the EXAFS spectra (see text). The decrease of pH accelerates the
carboplatin decomposition.
0.05 M HCl solutions, applying the same calculation for both
solutions. These results clearly show that the higher the
hydrochloric acid concentration, the faster the carboplatin
conversion.
NaCl carboplatin solutions
The fresh solutions (S and S ) show no visible evolution of
the spectrum between the Ðrst one and the last one; all spectra
are comparable to the carboplatin spectrum, either solid or in
Conclusion
7
8
This XAS study shows that this method is suitable for study-
ing the evolution of platinum-based antitumoral drugs in
solution, under various conditions, at least qualitatively. In
particular, we were able to determine the conditions under
which carboplatin is stable: room temperature, light exposure
or dark, aqueous medium solution, neutral pH, chloride con-
centration smaller than 0.9%, and use within a Ðfteen-days
delay.
When carboplatin conversion occurs, speciÐcally by addi-
tion of hydrochloric acid or under high chloride anion con-
centration, our XAS study demonstrates that cisplatin is
always the major degradation product. Although the evolu-
tion of the spectra is clear, the lack of homogeneity of the
solutions prevents us from determining the kinetics param-
eters of this carboplatin conversion.
solution, during the experimental time (7 h for S and 1.5 h for
S ).
7
8
After two weeks, only the 18% NaCl solutions (S5
and S )
d
5l
show an alteration: a yellow compound precipitates. The
spectrum of this compound is identical to the cisplatin spec-
trum. The remaining solution itself presents a very small
absorption edge (about 3% of the original absorption edge),
indicating that almost all the platinum is in the precipitate.
All other solutions (S to S ) showed no precipitate. Their
EXAFS spectra are comparable to the carboplatin solution
spectrum. These results show that the carboplatin is not
decomposed under these conditions, at the detection level of
XAS.
1
4
We plan to apply XAS spectroscopy to study other plati-
num drugs that are today under pharmaceutical evaluation
and which are expected to form homogeneous systems.
HCl carboplatin solutions
The spectra of the 0.1, 0.4 and 1 M HCl solutions showed a
rapid evolution during the recording time; this evolution is
accompanied by the formation of a yellow precipitate in the
solution. This evolution is evident in the FT, as Fig. 5 (with
References
1
M HCl) shows. The single peak characteristic of the carbo-
1
C. Valie
re, P. Arnaud, E. Caro†, J. F. Dauphin, G. Clement and
` 
F. Brion, Int. J. Pharm., 1996, 138, 125.
platin decreases, whereas a shoulder appears on the long path
length side of the peak. A comparison between these FTs and
the cisplatin FT shows that this shoulder appears at the same
distance as the part of the Ðrst peak of the cisplatin FT due to
the chloride atom contribution. For the 0.05 M solution, the
reaction is slower and just the beginning of the degradation is
observed.
2
J. A. HadÐeld, A. T. McGown, M. J. Dawson, N. Thatcher and
B. W. Fox, J. Pharm. Biomed. Anal., 1993, 167, 723.
3
4
W. J. F. van der Vijgh, Clin. Pharmacokinet., 1991, 21, 242.
J. Prat, M. Pujol, V. Girona, M. Munoz and L. A Sole, J. Pharm.
8 
Biomed. Anal., 1994, 12, 81.
5
6
M. Lederer and E. LeipzigÈPagani, Int. J. Pharm., 1998, 167, 223.
B. Benaji, T. Dine, M. Luyckx, C. Brunet, F. Goudaliez, M. L.
Mallevais, M. Cazin, B. Gressier and J. C. Cazin, J. Clin. Pharm.
T her., 1994, 19, 95.
For solution S1
2
, we waited for a few hours until the reac-
tion was complete, giving a yellow precipitate and a super-
natent solution the spectra of both the precipitate and the
supernatent solution can be superimposed onto the cisplatin
spectrum (Fig. 6). This Ñoating solution is evidently not
homogeneous and may contain solid Ðne particles of cisplatin.
Despite this lack of homogeneity, the precipitation of cisplatin
from a hydrochloric acid carboplatin solution is not question-
able.
7
8
Z. Guo, A. Habtemariam, P. J. Sadler, R. Palmer and B. S.
Potter, New J. Chem., 1998, 22, 11.
D. P. Bancroft, C. A. Lepre and S. J. Lippard, J. Am. Chem. Soc.,
1990, 112, 6860.
9
0
1
2
E. Curis, http://www.lure.u-psud.fr/LogicScient/LASE/lase.html.
1
1
1
E. Curis and S. Benazeth, J. Synchrotron Radiat., 2000, 7, 262.

G. H. W. Milburn and M. R. Truter, J. Chem. Soc. A, 1966, 1609.
B. Beagley, D. W. J. Cruickshank, C. A. McAuli†e, R. G. Prit-
chard and A. M. Zaki, J. Mol. Struct., 1985, 130, 97.
Qualitatively, it is possible to conÐrm the relatively quick
conversion of carboplatin into cisplatin with these HCl solu-
tions: since EXAFS oscillations s(k) are linearly additive, if a
is the molar fraction of cisplatin in solution and b that of
13 J. J. Rehr, J. Mustre de Leon, S. I. Zabinsky and R. C. Albers., J.
Am. Chem. Soc., 1991, 113, 5135.
1
1
1
4
5
6
P. N. V. Pavankumar, P. Seetharamulu, S. Yao, J. D. Saxe, D. G.
Reddy and F. H. Hausheer, J. Comput. Chem., 1999, 20, 365.
M. S. Lim and R. B. Martin, J. Inorg. Nucl. Chem., 1976, 38,
1911.
carboplatin, we must have for each spectrum s (k) \
exp
as
(k) ] bs (k), where s
(k) is the signal of cisplatin
S. A. Miller and D. A. House, Inorg. Chim. Acta, 1989, 166, 189.
CisPt
CbPt
CisPt
and s (k) is the signal of carboplatin. We used di†erent
17 S. A. Miller and D. A. House, Inorg. Chim. Acta, 1990, 173, 53.
CbPt
1008
New J. Chem., 2000, 24, 1003È1008