Mechanistic study of the oxidative degradation of the triazole antifungal agent CS-758 in an amorphous form

Article abstract of DOI:10.1002/jps.23339

In this study, the degradates generated from a pharmaceutical solid were characterized, and a mechanistic pathway underlying their formation was proposed. The chemical stability of a novel triazole antifungal drug, CS-758, deteriorated significantly when the crystal was disordered, and characteristic degradates were generated. A total of eight degradates in solution and nine degradates in a solid state were isolated by preparative liquid chromatography. Degradates were characterized using high-performance liquid chromatography-photodiode array, mass spectrometry, and nuclear magnetic resonance. Radical-mediated oxidation is proposed as the main degradation pathway in the solid state. The initiation step of this pathway is hydrogen atom abstraction from a methine carbon that is adjacent to a dien moiety and the formation of a delocalized vinylic radical intermediate. Molecular oxygen is then added to the radical position to form hydroperoxides. There are three potential oxidation routes based on the proposed autoxidation pathway that lead to the generation of the dioxane ring-opening hydroxyl form, the 9,10-epoxide form, or the 11,12-epoxide form, depending on the substituted position of the added molecular oxygen. The epimer compound generated via the vinylic radical intermediate and sulfoxides was characterized. This degradation mechanism provides the scientific foundation for an oxidative stressing system currently under investigation.

Full text of DOI:10.1002/jps.23339

Mechanistic Study of the Oxidative Degradation of the Triazole
Antifungal Agent CS-758 in an Amorphous Form
EIJI UEYAMA,1 NOBUYUKI SUZUKI,1 KENJI KANO2
1Analytical and Quality Evaluation Research Laboratories, Pharmaceutical Technology Division, Daiichi Sankyo,
Hiratsuka, Kanagawa 254-0014, Japan
2Division of Applied Life Sciences Graduate School of Agriculture, Kyoto University, Kitashirakawa Oiwake-cho,
Kyoto 606-8502, Japan
Received 12 June 2012; revised 13 September 2012;
Published online 5 October 2012 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.23339
ABSTRACT: In this study, the degradates generated from a pharmaceutical solid were char-
acterized, and a mechanistic pathway underlying their formation was proposed. The chemi-
cal stability of a novel triazole antifungal drug, CS-758, deteriorated significantly when the
crystal was disordered, and characteristic degradates were generated. A total of eight degra-
dates in solution and nine degradates in a solid state were isolated by preparative liquid
chromatography. Degradates were characterized using high-performance liquid chromatogra-
phy–photodiode array, mass spectrometry, and nuclear magnetic resonance. Radical-mediated
oxidation is proposed as the main degradation pathway in the solid state. The initiation step
of this pathway is hydrogen atom abstraction from a methine carbon that is adjacent to a
dien moiety and the formation of a delocalized vinylic radical intermediate. Molecular oxy-
gen is then added to the radical position to form hydroperoxides. There are three potential
oxidation routes based on the proposed autoxidation pathway that lead to the generation of
the dioxane ring-opening hydroxyl form, the 9,10-epoxide form, or the 11,12-epoxide form, de-
pending on the substituted position of the added molecular oxygen. The epimer compound
generated via the vinylic radical intermediate and sulfoxides was characterized. This degrada-
tion mechanism provides the scientific foundation for an oxidative stressing system currently
under investigation. © 2012 Wiley Periodicals, Inc. and the American Pharmacists Association
J Pharm Sci 102:104–113, 2013
Keywords: oxidation; amorphous; solid state; stability; chemical stability; degradation
products
the drug excipients3–5 or by the direct oxidation of
INTRODUCTION
the excipients themselves (e.g., the autoxidation of
In addition to hydrolysis, oxidation is a major path-
polysorbates).8,9 These oxidation reactions generally
way of drug degradation. Several different types of
involve complicated mechanisms. Oxidation reactions
oxidation mechanisms are known, such as autoxi-
sometimes generate unfavorable degradation prod-
dation (mediated by free radicals),1,2 nucleophilic/
ucts, such as hydroperoxide, aldehyde, and epoxide
electrophilic oxidation (mediated by peroxides),3,4
forms,10,11 which have structural features that may
electron transfer oxidation (mediated by transition
be linked to mutagenicity and carcinogenicity.12,13
metals),5 and photochemically induced oxidation (sin-
Therefore, oxidation is associated with a high risk
glet oxidation, etc.).6,7 In some cases, an oxidation
of generating potential genotoxic impurities.
reaction is triggered by trace amounts of impurities
An oxidative stressing system, which involves a
(e.g., peroxides or transition metals) contaminating
well-defined oxidation mechanism, can identify real-
istic oxidation products in advance and is useful for
determining the oxidation mechanism if the degra-
dation products correspond to those generated dur-
ing the storage of the actual drug. Furthermore,
the data obtained are helpful for developing strate-
gies to prevent or mitigate the risk of impurities,
Additional Supporting Information may be found in the online
version of this article. Supporting Information
Correspondence to: Eiji Ueyama (Telephone: +81-463-31-6888;
Fax: +81-463-38-3944; E-mail: ueyama.eiji.s2@daiichisankyo.co.jp)
Journal of Pharmaceutical Sciences, Vol. 102, 104–113 (2013)
© 2012 Wiley Periodicals, Inc. and the American Pharmacists Association
104
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 1, JANUARY 2013

MECHANISTIC STUDY OF THE OXIDATIVE DEGRADATION OF AMORPHOUS DRUG SUBSTANCE
105
such as improving the manufacturing process, for-
mulating an antioxidant, or packaging drugs with an
oxygen scavenger. Therefore, pharmaceutical compa-
nies have made efforts to develop oxidative stressing
systems that predict oxidation products that may be
generated during drug storage. Such stressing sys-
tems are used to evaluate drug candidates during the
early drug development stage.
The literature available regarding approaches for
studying oxidation mechanisms is extensive.14–20 In-
deed, these approaches have been used in the phar-
maceutical industry, and in many cases, the data
obtained about oxidative decomposition have been
helpful in the drug development process. However,
many of the oxidative stressing systems that have
been developed are solution based. Although such sys-
tems can be useful, they may not necessarily be reflec-
tive of the oxidation reactions that occur in the solid
state. Therefore, we are currently developing a solid-
based oxidative stressing system that can be applied
to the prediction of oxidation products generated in
pharmaceutical solids (manuscript in preparation).
Model drugs with known oxidation mechanisms are
necessary for the development of a solid-based oxida-
tive stressing system. A triazole antifungal agent, CS-
758, 4-[(1E,3E)-4-[trans-5-[[(1R,2R)-2-(2,4-di-fluorop-
henyl)-2-hydroxy-1-methyl-3-(1H-1,2,4-triazol-1-yl)
propyl]thio]-1,3-dioxan-2-yl]-1,3-butadienyl]-3-fluor-
obenzonitrile (Fig. 1), was the target compound in the
present study. This drug has a broad antifungal
spectrum covering Aspergillus sp. and fluconazole-
resistant Candida sp., as well as a good safety
profile.21 However, CS-758 shows chemical instabil-
ity when its crystal is disordered, and characteris-
tic oxidative degradates are generated. This report
describes investigations conducted to determine the
oxidative degradation pathway of one of the model
compounds used in a solid-based oxidative stressing
system that is under development.
Figure 1. Chemical structures of CS-758 (parent).
ple was cooled down by occasional refrigeration to
prevent melting. The powder X-ray diffraction pat-
tern of the ground sample showed a halo pattern,
and the disappearance of the diffraction peaks im-
plied that the drug substance had become amorphous
(data not shown). The purity of the amorphous drug
sample was greater than 98%, as determined by high-
performance liquid chromatography (HPLC) analysis.
Analytical HPLC Conditions
Agilent 1200 and 1100 systems (Agilent Technologies,
Waldbronn, Germany) equipped with a photodiode ar-
ray (PDA) detector were used for HPLC analysis. A
YMC Pack Pro C18 [150 × 3.0 mm2 internal diame-
ter (i.d.); particle size, 3 :m; YMC, Kyoto, Japan) was
used as an analytical column, and the temperature of
the column was kept at 40◦C. The eluent was mon-
itored at a wavelength of 220 nm. Mobile phases A
and B were 5 mM ammonium hydrogen carbonate and
acetonitrile, respectively. The gradient elution was
conducted as follows: from 0 to 12 min, the compo-
sition of mobile phase B was increased from 50% to
74%; from 12 to 16 min, its composition was increased
from 74% to 90%, and then maintained for 4 min at
a constant flow rate of 0.5 mL/min. The injection vol-
ume was 5 :L.
Preparative Liquid Chromatography Conditions
MATERIALS AND METHODS
Materials
Isolation of the degradates was performed using
an Agilent 1200 preparative liquid chromatography
(LC) system and an EZ-2 Plus solvent evaporator
(Genevac, Ipswich, UK). A YMC Pack Pro C18 (150 ×
10 mm2 i.d.; particle size, 5 :m; YMC) was used for
separation. Other chromatographic conditions were
the same as those used for analytical HPLC with the
following exceptions: (a) the flow rate was 5 mL/min,
(b) the injection volume was 900 :L, (c) the column
temperature was ambient, and (d) for the separation
of degradates SS-D6 and SS-D7, methanol was used
as mobile phase B.
CS-758 was provided by Daiichi-Sankyo (Tokyo,
Japan). All water used was purified using Milli-Q Gra-
dient A10 system (Millipore, Milford, Massachusetts).
R
The oxygen scavenger Sequlꢀ BP-100 was kindly do-
nated by NISSO JUSHI (Ibaraki, Japan). All other
chemicals were of analytical grade and from commer-
cial sources.
Preparation of Amorphous Sample using the Grinding
Method
Liquid Chromatography–Mass Spectrometry Conditions
The drug substance was ground using a vibra-
tional mill (RM-201; Mitsubishi Chemical Engineer-
ing, Tokyo, Japan) for a total of 120 min. The sam-
LC–mass spectrometry (LC–MS) and LC–tandem
mass spectrometry (LC–MS/MS) analyses were
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 1, JANUARY 2013

106
UEYAMA, SUZUKI, AND KANO
performed using an Agilent 1200 system equipped
with an LTQ Orbitrap XL mass spectrometer (Thermo
Fisher Scientific, Waltham, Massachusetts). The
HPLC conditions were the same as the analytical
HPLC conditions described previously. For online hy-
drogen–deuterium (H/D) exchange LC–MS analyses,
deuterium oxide was used as a solvent for the prepa-
ration of mobile phase A. Electrospray ionization in
positive polarity mode was utilized except in the case
of Acid-D2 detection, for which atmospheric pressure
chemical ionization in negative polarity mode was
used.
samples was confirmed. The compound identification
before, and following, isolation was conducted using
LC–PDA [retention time and online ultraviolet (UV)
spectrum], LC–MS (m/z), and LC–MS/MS (fragmen-
tation pattern) analyses.
Preparation of Amorphous Sample Degradates
Amorphous samples were placed in a headspace glass
vial that was closed with a cap. The sample was stored
in an oven at 50◦C for 4 weeks.
After the degradates were isolated by preparative
LC, isolated fractions were collected, combined, and
evaporated to obtain solid samples. A portion of the
solid samples was analyzed by the analytical HPLC
method, and the chromatographic purity of the sam-
ples was confirmed. The compound identification be-
fore, and following, isolation was conducted using
LC–PDA (retention time and online UV spectrum),
LC–MS (m/z), and LC–MS/MS (fragmentation pat-
tern) analyses.
We preliminarily confirmed that sample stability
was not easily affected by environmental humidity by
conducting comparative stress testing at 40◦C (close)
and 40◦C/75% RH (open). Because the effect of hu-
midity on the stability of samples was slight, humid-
ity was not considered as a factor in the storage of
samples during this study.
Nuclear Magnetic Resonance
Nuclear magnetic resonance (NMR) experiments
were conducted using an Avance 400 MHz NMR spec-
trometer (Bruker, Billerica, Massachusetts) equipped
with a broadband inverse 5 mm probe. All spec-
tra were acquired from isolated samples dissolved
in dimethylsulfoxide-d6 (DMSO-d6) or acetone-d6.
NMR experiments were performed at 25◦C. Chemi-
cal shifts are reported in parts per million (ppm) rela-
tive to DMSO-d6 (2.50 ppm) or acetone-d6 (2.05 ppm).
All spectra were recorded in spinning mode. Two-
dimensional techniques, including double-quantum
filter correlation spectroscopy (DQF-COSY), edited-
heteronuclear single-quantum correlation (edited-
HSQC) spectroscopy, and heteronuclear multiple-
bond correlation (HMBC) spectroscopy were used to
establish the H H network and the C H, C C H,
and C C C H connectivity, respectively.
Preparation of Antioxidant Mixture Samples by the
Solvent Evaporation Method
Preparation of Degradates in Solution
A total of 0.5 g of CS-758 and 0.005 g of antioxidants
such as butylated hydroxytoluene (BHT), propyl gal-
late, and ascorbic acid (100:1, w/w) were dissolved in a
mixture of acetone and ethanol (2:1, v/v). The solvent
was removed at 30◦C under reduced pressure using
a rotary evaporator for 2 h. The resultant dried mass
was crushed with a spatula, and the coarse-grained
powder was used as the solid sample. Using the same
method, a control sample was prepared by dissolving
0.5 g of CS-758 into the solvent mixture, then evap-
orating the solvent. These samples were then stored
for up to 2 weeks in a 50◦C oven. Aliquots of the sam-
ple were collected at specific intervals, dissolved in a
diluent, and analyzed using analytical HPLC.
Because the concentration of the drug in samples
changes over time as a result of both oxidation and
thermolysis, including the formation of other minor
degradation products, first-order kinetic theory is not
adequate for evaluating oxidative degradation in this
case. However, as the degradation apparently follows
first-order kinetic theory in this 2-week time range,
we adopted the theory to express relative degradation
rate numerically.
Approximately 10 mg of the drug was placed in a glass
bottle and dissolved in 8 mL of a dilute solution (a
mixture of acetonitrile and water 7:3, v/v). After 2 mL
of a 0.5 M HCl solution was added, the sample was
stored at 25◦C for 1 h. A second sample was prepared
in exactly the same manner, except that 2 mL of a
0.5 M NaOH solution was added instead of the 0.5 M
HCl solution, and the sample was stored at 25◦C for
24 h. A third sample was also prepared in exactly the
same manner, except that 2 mL of a 2.5% (w/w) H2O2
solution was added instead of the 0.5 M HCl solution,
and the sample was stored at 60◦C for 1 h. A fourth
sample was prepared by placing approximately 22 mg
of the drug and 6.5 mg of 2-2ꢀ-azobisisobutyronitrile
(AIBN) in a glass bottle, and dissolving the sample in
20 mL of acetonitrile. The solution was then stored at
40◦C for 1 week.
After the degradates were isolated using a prepar-
ative LC, isolated fractions were collected, combined,
and evaporated to obtain solid samples. A portion of
the solid samples was analyzed using the analytical
HPLC method and the chromatographic purity of the
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 1, JANUARY 2013
DOI 10.1002/jps

MECHANISTIC STUDY OF THE OXIDATIVE DEGRADATION OF AMORPHOUS DRUG SUBSTANCE
107
Figure 2. HPLC chromatograms (UV 220 nm) showing degradation profiles of CS-758
(a) overlaid chromatogram obtained from various forced degradation conditions in solution;
(i) 0.1 N HCl at 25◦C for 1 h, (ii) 0.1 N NaOH at 25◦C for 24 h, (iii) 0.5% (w/w) H2O2 at 60◦C for
1 h, (iv) 2 mM AIBN in acetonitrile at 40◦C for 1 week, and (b) chromatogram obtained from a
stressed amorphous (120 min grinded drug substance) under the condition at 50◦C for 3 weeks.
-D9, eluted at RRTs of 0.49, 0.59, 0.61, 0.64, 0.82, 0.84,
0.84, 1.58, and 1.64, respectively (Fig. 2b).
RESULTS
Profile of Degradates
High-performance liquid chromatography–PDA
and LC–MS data indicated that SS-D6 and SS-D7
were not separated in the preparative and analytical
HPLC conditions (Fig. 2b). Therefore, an alternative
HPLC method was investigated to separate SS-D6
and SS-D7. The mobile phase pH did not affect the
separation status; however, the addition of methanol
to the mobile phase resulted in better separation.
When mobile phase B was replaced with methanol,
SS-D6 and SS-D7 were completely separated with
a resolution of 3.4. To accomplish the isolation, SS-
D6 and SS-D7 were isolated as a mixture fraction
using the acetonitrile-based preparative LC method.
The mixture was then separated and isolated via the
methanol-based preparative LC method.
CS-758 was subjected to acidic, basic, and oxidative
conditions in solution. The data obtained from HPLC
analysis showed that eight major degradates were
generated under these stressing conditions (Fig. 2a).
Analytical HPLC showed that the acid degradates,
Acid-D1 and -D2, eluted at relative retention times
(RRTs) 0.16 and 0.42, respectively; the base degra-
dates, Base-D1 and -D2, eluted at RRTs 0.13 and 0.47,
respectively; the H2O2 degradates, HP-D1, -D2, and
-D3, eluted at RRTs 0.49, 0.59, and 0.77, respectively;
and the radical-initiator degradate RI-D1 eluted at
RRT 0.82.
The nine degradates generated in the solid state,
namely SS-D1, -D2, -D3, -D4, -D5, -D6, -D7, D-8, and
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 1, JANUARY 2013

108
UEYAMA, SUZUKI, AND KANO
Characterization of Degradates in Solution
The results of the LC–MS analyses are summarized
in Table 1. Accurate mass analysis of the degradates
demonstrated that a resolving power of 60,000 pro-
vided the best possible molecular formulas. These
formulas were within 5-ppm error of the theoretical
molecular weight. Structure information, including
estimated molecular formulae, number of active hy-
drogen atoms, MS/MS fragmentation (the proposed
MS/MS fragmentation of CS-758 and its degradates
are shown in Supporting Information-1), and λmax of
the online UV spectrum led to the proposal of reason-
able structures for these degradates (Fig. 3a).
Full assignment of 1H-NMR of RI-D1, the major ox-
idation product that was also generated in the solid
state, is provided in Supporting Information-2. NMR
indicated that RI-D1 was similar to the parent com-
pound except at positions 9, 10, 11, and 12, which
were located in the dien moiety. The proton–proton
correlation of the moiety was identical to that of the
parent compound, but the chemical shifts of the pro-
tons and carbons at positions 9 and 10 were drasti-
cally shifted to a high magnetic field (proton: 3.20 and
3.73 ppm, carbon: 59.1 and 54.7 ppm). This indicated
that the methine moiety changed from CH= to >CH.
The chemical shifts of the protons and carbons at posi-
tions 9 and 10 were consistent with the attachment of
oxygen atoms at these positions. Furthermore, accu-
rate mass data suggested that the molecular weight
of RI-D1 was 16 amu greater than CS-758, which cor-
responds to the weight of one additional oxygen atom.
Online H/D exchange LC–MS analysis of RI-D1 and
CS-768 indicated no difference in the number of ac-
tive hydrogen atoms in each compound. The online
UV–PDA spectrum demonstrated that the absorp-
tion maximum (λmax) of R1-D1 shifted to a shorter
wavelength relative to that of CS-768. These results
support the substitution of an oxygen at this posi-
tion, as shown in the epoxide structure proposed in
Figure 3a. Likewise, data from all analysis types sup-
ported the structures proposed for the other degra-
dates (full NMR spectrum assignments are provided
in Supporting Information-2). The proposed hydroly-
sis and oxidation pathways for the creation of these
degradates are shown in Figure 3a.
Characterization of Amorphous Sample Degradates
The degradation profile of the amorphous sample dif-
fered from those of the acidic and basic conditions
(Fig. 2). However, several of the degradates were con-
sistent with those generated under oxidative condi-
tions created by AIBN and H2O2 (RI-D1, HP-D1, and
HP-D2). SS-D3, SS-D4, SS-D6, SS-D7, SS-D8, and
SS-D9 could not be matched to degradates that were
generated in any of the forced solution-based degra-
dation conditions. The LC–MS results suggested that
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 1, JANUARY 2013
DOI 10.1002/jps

MECHANISTIC STUDY OF THE OXIDATIVE DEGRADATION OF AMORPHOUS DRUG SUBSTANCE
109
Figure 3. Chemical structures of degradates and degradation pathway in solution state.
(a) Main degradation pathway in acid, base, and oxidation condition in solution. (b) Acid-
catalyzed epimerization mechanism.
SS-D3, -D4, and -D7 might be simple oxidation prod-
ucts (parent + 16 amu), whereas SS-D6 appeared to
be an isomer of the parent. The molecular weights of
SS-D8 and -D9 indicated that they might be dimer
compounds (Table 1).
The LC–MS analyses and NMR data, including the
1H, DQF-COSY, edited-HSQC, and HMBC results,
also suggested that SS-D6 was an epimer of the par-
ent. Epimerization may occur at position 8 due to the
proposed autoxidation pathway (Fig. 4). However, it
was not possible to fully characterize the structure
of SS-D6 from these spectroscopic analyses alone.
Interestingly, this same degradate was also gener-
ated under acidic conditions as a minor degradate
(Fig. 2a, at 9.7 min). The main degradation pathway
that occurred under acidic conditions involved acid-
catalyzed dioxane ring-opening hydrolysis. In this
pathway, a hydroxyl group, which was generated via
the ring-opening reaction under acidic conditions, car-
ries out a nucleophilic attack on the carbocation at po-
sition 8 to form an epimer. Intramolecular recycliza-
tion may have occurred under acidic conditions. The
proposed acid-catalyzed epimerization mechanism is
shown in Figure 3b and the structure of the epimer
is supported from the viewpoint of the acid-catalyzed
epimerization reaction.
Given that SS-D3 and SS-D4 had the same spectro-
scopic properties, these degradates might be diastere-
omer compounds of each other. The NMR data showed
the disappearance of the methine proton resonance
at position 8, and the multiplicity of the resonance at
position 9 was changed from double doublet to dou-
blet in comparison with those of CS-768. HMBC data
showed that the protons at positions 7 (or 6 in the
case of its diastereomer) and 9 correlated with a char-
acteristic carbon downfield at 167 ppm. The chemical
shift of the carbon was attributed to a carbonyl car-
bon located in the adjacent positions at 7 (or 6) and
9. In addition, the edited-HSQC data indicated that
the two carbons at positions 6 and 7 were changed to
nonequivalent carbons. These results are supportive
of dioxane ring-opening hydroxyl forms (Fig. 4).
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 1, JANUARY 2013

110
UEYAMA, SUZUKI, AND KANO
Figure 4. Chemical structures of degradates and proposed oxidation pathway in solid state.
of oxygen at positions 11 and 12, as shown in the pro-
posed epoxide structure (Fig. 4).
The NMR data (Supporting Information-2) indi-
cated that SS-D7, like RI-D1, showed similarity in
structural resonances to the parent compound, except
at positions 9, 10, 11, and 12. The chemical shifts
of the protons and carbons at positions 11 and 12
were significantly shifted to high magnetic field in
comparison with those of CS-768. The LC–MS, online
UV–PDA, and NMR data supported the substitution
SS-D8 and SS-D9 (Fig. 5) were considered to be di-
astereomers of each other, as these degradates had
the same spectroscopic properties. The NMR data
showed similarity in the structural resonances be-
tween these compounds and the parent compound,
except at positions 9, 10, 11, and 12. In the case of
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 1, JANUARY 2013
DOI 10.1002/jps

MECHANISTIC STUDY OF THE OXIDATIVE DEGRADATION OF AMORPHOUS DRUG SUBSTANCE
111
Figure 5. Chemical structures of degradates and proposed thermolysis pathway in solid state.
SS-D8, DQF-COSY indicated key correlations of H-
9 (2.17−2.22 ppm) to H-9ꢀ (2.84 ppm), and of H-10
(3.08−3.14 ppm) to H-12ꢀ (4.09−4.11 ppm), through
H H connectivity; HMBC also showed reasonable key
correlations that supported the proposed structure.
The key correlations confirmed the presence of a six-
member cyclic ring. Furthermore, edited-HSQC data
confirmed that positions 11, 12, 10ꢀ, and 11ꢀ were part
of the methylidene group ( CH ). All of the correla-
tions observed in the DQF-COSY, edited-HSQC, and
HMBC experiments confirmed the proposed struc-
tures of SS-D8 and SS-D9.
form a delocalized vinylic radical intermediate, with
molecular oxygen being subsequently added to the
radical position to form hydroperoxides. On the basis
of the proposed autoxidation pathway, three oxidation
routes can be envisioned that lead to the generation
of the dioxane ring-opening hydroxyl forms (SS-D3
and -D4), the 9,10-epoxide form (SS-D5), or the 11,12-
epoxide form (SS-D7), depending on the substituted
position of molecular oxygen. Furthermore, the gener-
ation of the epimer compound (SS-D6) provides strong
support for the proposed initiation step. If the de-
localized vinylic radical intermediates are generated
before the addition of the oxygen molecule, the gener-
ation of 8,10-dien and 8,11-dien degradates would be
expected in addition to SS-D6 (9,11-dien). However,
8,10- and 8,11-dien degradates were not confirmed as
primary degradation products. Because it is difficult
for an 8-en structure to take a chair conformation, a
relatively sterically stable form of dioxane ring, it is
harder to generate than 9,11-dien.
DISCUSSION
Degradation Mechanism
Most degradation of amorphous samples likely occurs
via oxidation, as most of the degradates are oxidation
products. Proposed solid-state degradation pathways
that lead to the measured degradates are presented in
Figures 4 and 5. The oxidation pathway presented in
Figure 4 involves radical-mediated oxidation (autoxi-
dation), in which a hydrogen atom is abstracted from
the methine carbon adjacent to the dien moiety to
The proposed degradation mechanism was con-
firmed from two perspectives: (1) the stability of the
amorphous sample in an anaerobic condition and (2)
the effect of antioxidants on the rate of oxidation.
When the amorphous sample was stressed under an
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 1, JANUARY 2013

112
UEYAMA, SUZUKI, AND KANO
Table 2. Effect of Antioxidants on CS-758 Chemical Stability
Antioxidant
Mechanism
k (×103 day−1
)
Relative Rate Constant
Ascorbic acid
Propyl gallate
Butylated hydroxytoluene
Additive free (control)
Acid, preferentially oxidized
3.852
1.129
0.479
5.006
0.77
0.23
0.10
1.00
H-atom donor
H-atom donor
–
anaerobic condition that was created using an oxygen
scavenger, the generation of all oxidation degradates
was inhibited. The results of stressing the amorphous
compound by storage at 50◦C with or without an oxy-
gen scavenger for up to 4 weeks are shown in Sup-
porting Information-3. Furthermore, the addition of
antioxidants to the amorphous sample decreased the
oxidation rate (Table 2; the kinetic plot is shown in
Supporting Information-4). The effectiveness of the
antioxidants was ranked by the relative rate constant:
BHT > propyl gallate > ascorbic acid > control. These
results suggest that the CS-758 radical species medi-
ates the oxidation reaction in the amorphous sample
because BHT and propyl gallate, which act as radical
scavengers (H-atom donors), competitively prevented
oxidation.
In contrast, the generation of SS-D8 and SS-D9
was not prevented by the addition of any antioxi-
dants or by storage under an anaerobic condition.
This suggests that these degradates were not ox-
idation products, but products of thermolysis. We
proposed a Diels–Alder reaction as the chemical
degradation pathway (Fig. 5), with the generation
rate dependent on temperature. This reaction mecha-
nism provides a possible explanation for the empirical
observation that the generation of SS-D8 and SS-D9
was not prevented by the addition of antioxidants or
by storage under an anaerobic condition. In addition,
the LC–MS results showed that two minor peaks,
which were eluted in the range of 17–18 min on the
chromatogram shown in Figure 2, have the same m/z
as SS-D8 and SS-D9. The molecular weights of these
minor peaks indicated that they might be other dimer
compounds. Although these minor peaks were not iso-
lated, these compounds may be Diels–Alder reaction
products of the C11–C12 double bond with the 9,11-
dien moiety. There may have been a sterical restric-
tion that prevented large quantities of these products
from forming.
state, were characterized. Radical-mediated oxida-
tion (autoxidation) was proposed as the main degra-
dation pathway in the solid state. The initiation step
of this oxidation reaction was hydrogen atom abstrac-
tion from the methine carbon adjacent to the dien moi-
ety to form a delocalized vinylic radical intermediate.
Molecular oxygen was subsequently added to the rad-
ical position to form hydroperoxides. On the basis of
the proposed autoxidation pathway, three oxidation
routes were envisioned that lead to the generation of
characteristic oxidation products, depending on the
substituted position of molecular oxygen. Further-
more, the generation of the epimer compound (SS-D6)
provided strong support for the proposed initiation
step. The H-atom donor-type antioxidants, BHT, and
propyl gallate, as well as the oxygen scavenger, sig-
nificantly reduced the rate of oxidation, suggesting
that the drug radical species mediates the oxidation
reaction in the amorphous form (autoxidation). The
knowledge acquired on the degradation pathway of
this model compound is useful for the development
of an oxidative stressing system, which is currently
under investigation.
ACKNOWLEDGMENTS
The authors would like to thank Daiichi Sankyo
Company for encouraging publication of this work.
REFERENCES
1. Baertschi SW, Brunner H, Bunnell CA, Cooke GG, Diseroad
B, Dorman DE, Jansen PJ, Kemp CA, Maple SR, McCune KA,
Speakman JL. 2008. Isolation, identification, and synthesis of
two oxidative degradation products of olanzapine (LY170053)
in solid oral formulations. J Pharm Sci 97:883–892.
2. Ueyama E, Takahashi F, Ohashi J, Konse T, Kishi N, Kano
K. 2012. Mechanistic study on degradation of azelnidipine
solution under radical initiator-based oxidative conditions. J
Pharm Biomed Anal 61:277–283.
3. Dyakonov T, Muir A, Nasri H, Toops D, Fatmi A. 2010. Iso-
lation and characterization of cetirizine degradation product:
Mechanism of cetirizine oxidation. Pharm Res 27:1318–1324.
4. Hartauer KJ, Arbuthnot GN, Baertschi SW, Johnson RA, Luke
WD, Pearson NG, Rickard EC, Tingle CA, Tsang PK, Wiens
RE. 2000. Influence of peroxide impurities in povidone and
crospovidone on the stability of raloxifene hydrochloride in
tablets: Identification and control of an oxidative degradation
product. Pharm Dev Technol 5:303–310.
5. Hong J, Lee E, Carter JC, Masse JA, Oksanen DA. 2004.
Antioxidant-accelerated oxidative degradation: A case study of
transition metal ion catalyzed oxidation in formulation. Pharm
Dev Technol 9:171–179.
In the initiation step of the proposed oxidation
mechanism (Fig. 4), the hydrogen atom at position
8 is abstracted by some foreign element. Although
the true foreign element that initiates the autoxida-
tion remains unknown, heat, trace impurities (e.g.,
radical initiators, peroxides, or transition metals), or
mechanoradicals generated by mechanical forces22 is
possible initiator elements.
In conclusion, the degradates of the triazole anti-
fungal agent CS-758, both in solution and in solid
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 1, JANUARY 2013
DOI 10.1002/jps

MECHANISTIC STUDY OF THE OXIDATIVE DEGRADATION OF AMORPHOUS DRUG SUBSTANCE
113
6. Seburg RA, Ballard JM, Hwang TL, Sullivan CM. 2006. Pho-
tosensitized degradation of losartan potassium in an extem-
poraneous suspension formulation. J Pharm Biomed Anal
42:411–422.
7. Qin X, Frech P. 2001. Liquid chromatography/mass spectrom-
etry (LC/MS) identification of photooxidative degradates of
crystalline and amorphous MK-912. J Pharm Sci 90:833–844.
8. Yao J, Dokuru DK, Noestheden M, Park SS, Kerwin BA, Jona
J, Ostovic D, Reid DL. 2009. A quantitative kinetic study of
polysorbate autoxidation: The role of unsaturated fatty acid
ester substituents. Pharm Res 26:2303–2313.
9. Kerwin BA. 2008. Polysorbates 20 and 80 used in the formu-
lation of protein biotherapeutics: Structure and degradation
pathways. J Pharm Sci 97:2924–2935.
10. Harmon PA, Biffar S, Pitzenberger SM, Reed RA. 2005. Mech-
anism of the solution oxidation of refecoxib under alkaline
conditions. Pharm Res 22:1716–1726.
11. Boccardi G. 2005. Oxidative susceptibility testing. In Pharma-
ceutical stress testing: Predicting drug degradation; Baertschi
SW, Ed. New York City, New York: Taylor & Francis, pp
205–234.
12. Ball JC, Young WC, Wallington TJ, Japar SM. 1992. Muta-
genic properties of a series of alkyl hydroperoxides. Environ
Sci Technol 26:397–399.
13. Muller L, Mauthe RJ, Riley CM, Andino MM, Antonis DD,
Beels C, DeGeorge J, De Knaep AG, Ellison D, Fagerland
JA, Frank R, Fritschel B, Galloway S, Harpur E, Humfrey
CD, Jacks AS, Jagota N, Mackinnon J, Mohan G, Ness DK,
O’Donovan MR, Smith MD, Vudathala G, Yotti L. 2006. A ra-
tionale for determining, testing, and controlling specific impu-
rities in pharmaceuticals that possess potential for genotoxic-
ity. Regul Toxicol Pharmacol 44:198–211.
15. Reynolds DW, Galvani M, Hicks SR, Joshi BJ, Kennedy-
Gabb SA, Kleinman MH, Parmar PZ. 2012. The use
of N-methylpyrrolidone as
pharmaceutical stress testing.
776.
a
cosolvent and oxidant in
Pharm Sci 101:761–
J
16. Kawaguchi-Murakami Y, Fukutsu N, Kajiro T, Araki T, Mu-
rakami T, Kawasaki T, Kishi N, Konno M. 2009. A prediction
system of oxidation reaction as a solid-state stress condition:
Applied to a pyrrole-containing pharmaceutical compound. J
Pharm Biomed Anal 50:328–335.
17. Nelson ED, Harmon PA, Szymanik RC, Teresk MG, Li L,
Suburg RA, Reed RA. 2006. Evaluation of solution oxygena-
tion requirements for azonitrile-based oxidative forced degra-
dation studies of pharmaceutical compounds. J Pharm Sci
95:1527–1539.
18. Harmon PA, Kosuda K, Nelson E, Mowery M, Reed RA. 2006.
A novel peroxy radical based oxidative stressing system for
ranking the oxidizability of drug substances. J Pharm Sci
95:2014–2028.
19. Betiqeri S, Thakur A, Raqhavan K. 2005. Use of 2,2ꢀ-azobis(2-
amidinopropane) dihydrochloride as a reagent tool for eval-
uation of oxidative stability of drugs. Pharm Res 22:310–
317.
20. Reid DL, Calvitt CJ, Zell MT, Miller KG, Kingsmill CA. 2004.
Early prediction of pharmaceutical oxidation pathways by
computational chemistry and forced degradation. Pharm Res
21:1708–1717.
21. Kamai Y, Harasaki T, Fukuoka T, Ohya S, Uchida K,
Yamaguchi H, Kuwahara S. 2002. In vitro and in vivo activ-
ities of CS-758 (R-120758), a new triazole antifungal agent.
Antimicrob Agents Chemother 46:367–370.
22. Polizzi MA, Singhal D, Colvin J. 2008. Mechanoradical-
14. Zhu DA, Zhang GG, George KL, Zhou D. 2011. A novel acceler-
ated oxidative stability screening method for pharmaceutical
solids. J Pharm Sci 100:3529–3538.
induced degradation in
ing high-shear processing. Pharm Dev Technol 13:457–
462.
a
pharmaceutical blend dur-
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 1, JANUARY 2013