Rate and equilibrium constants for the epimerization of the endothelin receptor antagonist J-104,132 in aqueous solution

Article abstract of DOI:10.1248/cpb.49.1

The degradation of [5S-[5α,6β,7α(R*)]]-2-butyl-5-(1,3-benzodioxol-5-yl)-7 -[(2-carboxypropyl)-4-methoxyphenyl]-6-dihydro-5H-cyclopenta[b]pyridine-6-car boxylic acid (J-104,132) was studied in aqueous solution as a function of temperature and pH. The degra

Full text of DOI:10.1248/cpb.49.1

January 2001
Chem. Pharm. Bull. 49(1) 1—4 (2001)
1
Rate and Equilibrium Constants for the Epimerization of the Endothelin
Receptor Antagonist J-104,132 in Aqueous Solution
a
a
a
,a
*
Michelle L. BRAY, Daniela GORBACHEVA, Hossain JAHANSOUZ, Michael J. KAUFMAN,
Kiyofumi ISHIKAWA,b Nobuo HARADA,b and Kimimasa SUZUKI
b
Pharmaceutical Research and Development, Merck Research Laboratories,a WP78A-31, West Point, PA 19486, U.S.A. and
Pharmaceutical Research and Development, Laboratories for Technology Development, Banyu Pharmaceutical Co., Ltd.,b
810, Nishijo, Menuma-Machi, Osato-Gun, Saitama 360–0214, Japan.
Received September 13, 1999; accepted October 23, 2000
The degradation of [5S-[5a,6b,7a(R*)]]-2-butyl-5-(1,3-benzodioxol-5-yl)-7-[(2-carboxypropyl)-4-methoxy-
phenyl]-6-dihydro-5H-cyclopenta[b]pyridine-6-carboxylic acid (J-104,132) was studied in aqueous solution as a
function of temperature and pH. The degradation reaction does not proceed to completion; rather, a stable equi-
librium is attained in which approximately 2% of the degradate is produced. Kinetic data for the formation of
the degradate are analyzed using an integrated form of the rate law for a reversible first-order reaction, and the
forward and reverse rate constants and overall equilibrium constants are presented. Isolation and spectroscopic
structural determination indicate that the degradate is the C7 b-epimer of the drug. A mechanism for the
epimerization reaction involving a novel enamine-like intermediate is proposed and shown to be consistent with
the kinetic data. The rate and equilibrium constants are used to predict the room temperature stability of an in-
jectable formulation of J-104,132, and these predictions are compared to actual data from long-term stability
studies. It is concluded that the preformulation kinetic studies provide essential data needed for optimum drug
product development.
Key words stability; kinetics; equilibrium; epimerization; endothelin receptor antagonist; preformulation
An essential part of the pharmaceutical development of in terms of forward and reverse rate constants for epimeriza-
new drugs is the quantitative determination of their chemical tion. Finally, the accelerated stability data are used to derive
reactivity under conditions that are relevant to the way they the room temperature reactivity of the drug in a probe intra-
are manufactured, stored, and used. The goals of preformu- venous formulation, and these predictions are compared to
lation stability studies are two-fold. First, it is important to long-term data obtained at 25 °C.
obtain detailed kinetic information under a variety of envi-
ronmental conditions so that the storage shelf life of a new
drug can be estimated. Second, the mechanism of degrada-
Experimental
Materials All solvents were obtained from Fisher Scientific (Pittsburgh,
PA, U.S.A.) and were HPLC grade or equivalent. Reagents were obtained
tion must be determined so that the formulation can be scien-
tifically designed to maximize the stability of the active in-
gredient in the drug product. Due to the generally higher re-
activity of drugs in solution as opposed to the solid phase,
preformulation studies are particularly important for com-
pounds that will be formulated as aqueous solutions.
The relatively low reactivity of many substances makes it
necessary to conduct high temperature stability studies and
extrapolate the results to room temperature. This procedure,
however, may be prone to gross inaccuracies under circum-
stances where the observed reactivity is the sum of several
reaction steps such as oxidation reactions.1) Also, extrapola-
tion of high temperature data will lead to errors if there is a
change in the reaction mechanism over the temperature range
of interest. Thus, a detailed analysis of kinetic, mechanistic,
from Fisher Scientific and were certified ACS grade. Laboratory samples of
J-104,132 were obtained from Banyu Pharmaceutical Co. Ltd. (Menuma,
Japan); the samples were greater than 99% pure as judged by reverse phase
HPLC and were used without further purification.
Kinetic Procedures Solutions used for kinetic studies consisted of the
drug (5.0 mg/ml) and mannitol (42 mg/ml, used to provide physiologically
isotonic media) dissolved in a 0.010 M buffer. The buffers used were sodium
citrate/hydrochloric acid (pH 6.5 and 7.5) and tris/hydrochloric acid (pH
8.5). Measurement of pH was accomplished with a Fisher Accumet Model
15 pH meter.
The kinetic runs were all performed in essentially the same manner, and a
typical run is described below:
A solution of the drug in the mannitol/buffer vehicle was adjusted with
0.01 M hydrochloric acid to the desired pH (Ϯ0.1 pH unit), then filtered
through a 0.22 mm disc filter. The filtrate was filled in 2 ml portions into
glass ampoules, and the ampoules were flame sealed. Samples were stored in
constant temperature ovens (Thermolyne Series 9000) at 40, 60 and 80 °C
(Ϯ0.5 °C). At recorded time intervals, samples were removed from storage,
and product data are needed to obtain reliable predictions of allowed to come to room temperature, then quantitatively diluted with deion-
room temperature stability for many pharmaceuticals. In this
paper, we present the results of preformulation stability stud-
ies with [5S-[5a,6b,7a(R*)]]-2-butyl-5-(1,3-benzodioxol-5-
yl)-7-[(2-carboxypropyl)-4-methoxyphenyl]-6-dihydro-5H-
cyclopenta[b]pyridine-6-carboxylic acid (J-104,132, Fig. 1),
a potent and specific non-peptidyl antagonist for the endothe-
lin receptor.2) Kinetic and product studies in aqueous solution
indicate that J-104,132 is prone to an acid-catalyzed epimer-
ization reaction, and we present a mechanism that involves a
novel enamine intermediate. Studies of the reaction as a
function of pH and temperature are presented and analyzed Fig. 1. Chemical Structure of J-104,132
To whom correspondence should be addressed. e-mail: Michael_Kaufman@Merck.com
© 2001 Pharmaceutical Society of Japan

2
Vol. 49, No. 1
ized water and assayed for intact drug and degradate levels by HPLC. HPLC
analyses were performed on a Waters HPLC system (Model 600 controller
and 717 Plus autoinjector equipped with a Model 486 UV detector) using a
YMC ODS-AM column (4.6ϫ250 mm) with UV detection at 230 nm. The
compounds were eluted with a linear gradient of 40 to 90% acetonitrile vs.
0.1% aqueous phosphoric acid over 30 min. The drug elutes with a retention
time of 15.4 min, and the degradate elutes at 11.2 min.
In separate experiments, it was shown that the HPLC response factor for
J-104,132 and its degradation product are identical to within Ϯ10%. Thus,
all calculations of rate and equilibrium constants were performed using
HPLC area percent values taken directly from the chromatograms.
Degradate Isolation The chemical structure of the degradation product
of J-104,132 was determined by spectroscopic methods following the isola-
tion of an analytically pure sample. The degradate was isolated from stressed
(80 °C/2 weeks) samples of J-104,132 in 0.1 N hydrochloric acid using
preparative scale HPLC. The preparative separation was accomplished on a
Waters PrepLC 4000 instrument using an Inertsil Prep-ODS column
(250ϫ50 mm) and a mobile phase consisting of 0.1% aqueous trifluo-
roacetic acid and acetonitrile. Using a linear gradient from 30 to 70% ace-
tonitrile over 90 min at a flow rate of 25 ml/min, the degradate elutes at
42.0 min followed by the intact parent compound at 48.4 min. A sample
loading of 25 mg (5 mlϫ5 mg/ml) per injection was employed. The degra-
date fractions from 10 injections were combined, most of the acetonitrile and
volatile trifluoroacetic acid was removed on a rotary evaporator, and the re-
maining solution was lyophilized to provide an off-white solid that was
shown to be 98.3% pure by analytical HPLC. The major impurity in the iso-
lated sample was J-104,132.
Fig. 2. Plot of Degradate Level (HPLC Area%) vs. Time for a Kinetic Run
at pH 7.5 and 80 °C
Table 1. Rate and Equilibrium Constants for J-104,132 Degradation
Entry T (°C) pH 102k (dϪ1
)
102Keq 102kf (dϪ1
)
102kr (dϪ1
)
Proton NMR Spectroscopy 1H-NMR spectra for J-104,132 and the iso-
lated degradate were acquired on a Varian Unity-500 NMR spectrometer op-
erating at 500.154 MHz. The samples for NMR measurement contained ap-
proximately 1 mg of solid dissolved in 0.6 ml CD3CN. All spectra were ac-
quired at 25 °C. Proton chemical shift assignments were made with the com-
bined use of 1D proton, Nuclear Overhauser Effect (NOE) difference3,4) and
COSY experiments to determine the structure of the isolated degradate.
Mass Spectrometry The fast atom bombardment mass spectra of J-
104,132 and the degradate were obtained on a VG-70E magnetic sector mass
spectrometer operating at an accelerating potential of 6 kV. The analyte (ca.
1 mg) was dissolved in methanol and diluted with dithiothreotol/dithioery-
thritol (magic bullet) matrix in methanol. Magnetic scans were performed
over the mass/charge range of 800 to 300 at a scan rate of 6 s/decade.
UV Spectroscopy Ultraviolet spectra were recorded on a Perkin Elmer
Lambda 6 spectrophotometer. The samples were dissolved in 50/50 acetoni-
trile/water at a concentration of ca. 15 mg/ml and spectra were recorded from
190 to 350 nm at ambient temperature.
1
2
3
40
60
80
60
80
60
60
60
7.5
7.5
7.5
7.5
7.5
5.5
6.5
8.5
0.965
12.7
1.75
2.23
2.56
2.40
2.47
2.53
2.34
2.15
0.0166
0.277
2.77
0.948
12.4
111
108
4a)
5a)
6
18.5b)
ndc)
0.433b)
ndc)
18.0b)
ndc)
46.8
24.0
6.76
1.15
0.549
0.142
45.6
23.5
6.62
7
8
a) Degradate was the starting material for this run. b) Approximate value based on
limited kinetic data. c) nd, not determined.
104,132 as the reactant. This demonstrates that the final con-
centration of reactant and product is independent of the start-
ing material, and confirms the equilibrium nature of the reac-
tion.
The kinetic data for the formation of the degradate was an-
alyzed using the integrated form of the rate law for a re-
versible unimolecular reaction:5)
Results and Discussion
Decomposition Kinetics Buffered aqueous solutions of
J-104,132 slowly decompose over time to give a single reac-
tion product. The rate of decomposition is a function of the
solution temperature and pH, but is not effected by the initial
drug concentration. The reaction does not proceed to com-
pletion, but rather appears to reach a stable equilibrium state
in which approximately 2—3% of the degradate and 97—
98% of J-104,132 are present. This is shown graphically in
Fig. 2, where the degradate level is plotted as a function of
time for a kinetic run at pH 7.5. In particular, the data ob-
tained at 80 °C show that equilibrium is reached in ca. 96 h,
after which no further change in the level of drug or degra-
date is observed.
kf
Drug
Degradate
kr
[Degradate]ϭ[Degradate]eq·(1ϪeϪkt
)
(1)
where [Degradate] and [Degradate]eq represent the concen-
tration of the degradate at time t and at equilibrium, respec-
tively, and k is the observed rate constant for the reaction.
The relationship between the overall observed rate constant
and the elementary rate constants for the forward reaction kf
and the reverse reaction kr is given by Eq. 2. Additionally, the
equilibrium constant for the reaction is given by Eq. 3.
kϭkfϩkr
Keqϭkf/kr
(2)
(3)
Further experimental evidence that an equilibrium exists
between the drug and the degradate is provided by kinetic
runs performed using the degradate as the starting material.
The pure degradate (see next section) was dissolved in the
mannitol/buffer vehicle at pH 7.5, stored at 60 and 80 °C, and
analyzed by HPLC at measured time intervals. At both tem-
peratures, the only product observed following incubation of
the degradate under these conditions was J-104,132. More-
over, the final levels of drug and degradate were observed to
be the same as found in the kinetic runs starting with J-
In practice, the [Degradate] vs. time data were fitted to Eq.
1 by non-linear least squares techniques6) using [Degradate]eq
and k as adjustable parameters. The equilibrium constant was
obtained from the best fit estimate of [Degradate]eq, and the
forward and reverse rate constants were obtained by simulta-
neous solution of Eqs. 2 and 3. A summary of the kinetic re-
sults obtained in this fashion is shown in Table 1.

January 2001
3
Fig. 5. Degradation of J-104,132 to Form the C7 b-Epimer Degradate
One-dimensional proton difference NOE experiments of
Fig. 3. Temperature Dependence of the Equilibrium Constant for J- the degradation product with irradiation of H-6 (3.85 ppm)
104,132 Degradation
resulted in a strong NOE in H-7 (5.32 ppm) and a weak NOE
in H-5 (5.04 ppm), indicating that H-6 and H-7 are cis and H-
5 and H-6 are trans. Strong NOEs were observed in the 6.75
and 6.82 ppm dioxaindane protons whereas no NOE was ob-
served in the 6.57 ppm methoxyphenyl proton. In addition, a
strong NOE was observed in the 6.57 ppm proton upon satu-
ration of H-5 indicating that H-5 and the methoxyphenyl
group are in a cis relationship. In the case of J-104,132, only
a weak NOE was observed between H-5 and H-6, and H-6
and H-7, indicating that H-5 and H-6 and H-6 and H-7 are
trans. In addition, the corresponding NOE observed in the
6.57 ppm proton upon saturation of H-5 was absent for J-
104,132. Thus, the NMR data combined with the UV and
MS results indicate that the degradate is the C-7 b-epimer of
Fig. 4. Temperature Dependence of the Forward (᭿) and Reverse (᭹)
Rate Constants for J-104,132 Degradation
the parent drug as shown in Fig. 5.
As discussed in the previous section, the low value of DS
for the reaction is consistent with an epimerization reaction
that does not change the overall number of reactant and prod-
uct molecules. The equilibrium is essentially dictated by the
higher free energy of the b-isomer. This is presumably the re-
sult of an additional gauche interaction between the C6 and
C7 substituents that are cis to each other in the b-epimer, but
are in the trans configuration in the a-epimer. In this regard,
it is interesting that the 2.1 kcal/mol energy difference deter-
mined in our studies is quite close to the measured value of
1.7 kcal/mol between the cis and trans isomers of 1,2-di-
methylcyclopentane.7)
Aside from the well-known mutarotation of sugars,
epimerization is a relatively rare mechanism for the degrada-
tion of pharmaceutical agents. Ergotamine has been reported
to undergo acid catalyzed epimerization,8) and the activation
energy of 24.3 kcal/mol for that process9) is similar to the
value we have determined. Similarly, scopolamine undegoes
epimerization at the benzylic site in the molecule; an activa-
tion energy of 20.4 kcal/mol was reported for the base-cat-
alyzed process.10)
Mechanism of Epimerization The data presented in en-
tries 2, 6, and 7 of Table 1 indicate that the rate of epimeriza-
tion is accelerated as the solutions become more acidic, but
the extent of epimerization is unaffected by pH. This sug-
gests that the reaction is an acid catalyzed process. Conse-
quently, the mechanism of epimerization is proposed as
shown in Fig. 6. The key steps in the sequence are: (1) proto-
nation of the pyridine nitrogen to activate the C7 hydrogen,
(2) deprotonation to form an enamine intermediate with at-
tendant loss of chirality at C7, and (3) reprotonation to gen-
erate the product. In this sequence, the formation of the
The reactions at pH 7.5 were conducted at three tempera-
tures, thus affording the opportunity to obtain additional
thermodynamic information. The equilibrium constants were
plotted as ln Keq vs. inverse absolute temperature (Fig. 3),
and the derived enthalpy and entropy of reaction are
DHϭ2.1 kcal/mol and DSϭ1.3 cal/mol·K. Note that the
degradation reaction is enthalpically disfavored but entropi-
cally favored. However, the magnitude of the entropy change
is quite small and the enthalpy term dominates at tempera-
tures at or near room temperature. It is also notable that the
entropy of reaction is quite close to zero, which suggests that
the reaction proceeds with no change in the number of prod-
ucts relative to reactants.
The temperature dependence of the forward and reverse
rate constants for the reaction were observed to obey the Ar-
rhenius law as shown in Fig. 4. The activation energies for
the forward and reverse reactions are 28.1 kcal/mol and
26.0 kcal/mol, respectively. Thus, although the difference in
free energy between the reactant and product molecules is
only 2.1 kcal/mol, the kinetic barrier to the reaction is sub-
stantially higher.
Degradate Identification The degradation product was
isolated by preparative HPLC in quantities sufficient for
spectroscopic structure determination. The UV spectrum of
the degradate exhibits absorption maxima at 201, 230 and
281 nm, and is identical to that of the parent compound.
Mass spectral data indicate no change in the molecular mass
for the degradation product relative to the intact drug. Both
samples have a molecular mass of 531.3 by mass spec-
troscopy. The degradate is therefore an isomer of J-104,132.

4
Vol. 49, No. 1
Shelf Life Estimates The kinetic and equilibrium data
obtained in the accelerated stability studies can be used to
predict the stability of the drug under common storage condi-
tions. For example, the calculated rate and equilibrium con-
stants at pH 7.5 are kϭ1.20ϫ10Ϫ3 dϪ1 and Keqϭ0.015 at
25 °C, and kϭ5.12ϫ10Ϫ5 dϪ1 and Keqϭ0.0116 at 5 °C. If the
amount of degradate that can be tolerated in a pharmaceuti-
cal formulation is considered to be 0.5%,12) then the data in-
dicate that this threshold will be surpassed after 49 weeks of
storage at 25 °C. However, due to the large activation energy
of the reaction, the temperature dependence is large and the
degradate threshold for a formulation stored under refriger-
ated (5 °C) conditions will not be surpassed for over 30
years.
Subsequent to the preformulation studies described herein,
J-104,132 was formulated as a sterile injectable solution con-
taining mannitol and citrate buffer at pH 7.5. Long term sta-
bility data at 5, 25, and 30 °C were obtained to support the
continued use of the product in clinical trials. As predicted,
samples stored at 5 °C (the actual storage condition) showed
no detectable epimer levels after 1 year. The actual data ob-
tained in the stability studies at 25 and 30 °C are shown in
Table 2 along with the values predicted from the preformula-
tion studies. It can be seen that the agreement between pre-
dicted and actual values is very satisfactory. Thus, the degra-
dation of J-104,132 provides an excellent example of how
precise physicochemical analyses can be used to support and
optimize pharmaceutical product development.
Acknowledgements The authors would like to acknowledge Drs. San-
dor Varga and Harry Ramjit for providing NMR and mass spectral data, re-
spectively. Dr. Joseph Bondi was a great assistance in helping to coordinate
the transfer of information between the Merck and Banyu laboratories. We
also thank Dr. Orn Almarsson for proof-reading the text and providing help-
ful suggestions.
Fig. 6. Proposed Mechanism for the Acid-Catalyzed Epimerization Reac-
tion
Table 2. Observed and Predicted Levels of the C7-Epimer in an Injectable
Formulation
T (°C)
Time (weeks) % Epimer, observed % Epimer, predicted
References and Notes
1) Kaufman M. J., Pharm. Res., 7, 289—292 (1990).
2) Nishikibe M., Ohta H., Okada M., Ishikawa K., Hayama T., Fukuroda
T., Noguchi K., Saito M., Kanoh T., Ozaki S., Kamei T., Hara K.,
William D., Kivlighn S., Krause S., Gabel R., Zingaro G., Nolan N.,
O’Brien J., Clayton F., Lynch J., Pettibone D., Siegl P., J. Pharmacol.
Exp. Ther., 289, 1262—1270 (1999).
3) Braunschweiler L., Ernst R. R., J. Magn. Reson., 53, 521—525 (1983).
4) Bax A., Davis D. G., J. Magn. Reson., 65, 355—360 (1995).
5) Moore J. W., Pearson R. G., “Kinetics and Mechanism,” 3rd Edition,
Wiley-Interscience, New York, 1981. Note that, since the amount of
degradate present at tϭ0 in the starting material was essentially zero
(below limit of detection), we have used a simplified form of the equa-
tions cited in this reference.
25
4
13
26
39
52
4
Ͻ0.1
0.1
0.2
0.3
0.4
Ͻ0.1
0.2
0.5
0.05
0.15
0.29
0.41
0.52
0.11
0.32
0.57
30
13
26
enamine intermediate is undoubtedly the rate-determining
step in the reaction, and the acidifying effects of the adjacent
pyridinium and aryl groups are required for this step to
occur. Note that the effect of solution pH on the rate is com-
plex since the first step (protonation of the pyridine) is fa-
vored by acid but the rate-determining formation of the
enamine requires C7 to act as a carbon acid and is thus base-
promoted. Thus, reaction rates are expected to increase with
decreasing pH, but less so than the usual ten-fold increase in
rate for every 1 unit decrease in pH.11) This is consistent with
the kinetic data.
6) Curve fitting was performed with KaliedaGraphTM, Version 3.09.
7) Haresnape J. N., Chem. Ind., 1953, 1091—1100.
8) Ott H., Hofmann A., Frey A. J., J. Am. Chem. Soc., 88, 1251—1259
(1966).
9) Kreilgard B., Kisbye J., Arch. Pharm. Chem. Sci. Ed., 2, 38—56
(1974).
10) Smithius L. O. M. J., Pharm. Weekbl., 104, 1097—1104 (1969).
11) van der Houwen O. A. G. J., de Loos M. R., Beijnen J. H., Bult A., Un-
derberg W. J. M., Int. J. Pharm., 155, 137—152 (1997).
12) This value is based on current ICH guidelines as set forth in Q3B: Im-
purities in New Drug Products, November 1996.