Elucidating the pathways of degradation of denagliptin

Article abstract of DOI:10.1002/jps.22069

Stress testing or forced degradation studies of denagliptin (1) tosylate in solution and solid-state, its blends with excipients, and capsules were conducted in order to elucidate degradation pathways, aid formulation development, and generate data to support regulatory filings. In solution, denagliptin was stressed in acid, water, and base using organic cosolvents. In the solid-state, denagliptin was stressed under heat, humidity, and light. Blends of denagliptin with various excipients were stressed under heat and humidity in order to evaluate whether tablet was a viable dosage form. Capsules were stressed under heat, humidity, and light. It was found that denagliptin was stable in the solid-state, but degraded in solution, in blends with all excipients, and in capsules predominantly by cyclization to (3S,7S,8aS) amidine (2), which epimerized to (3S,7S,8aR) amidine (3). (3S,7S,8aR) amidine (3) subsequently hydrolyzed to the corresponding diketopiperazine (4). The purpose of this manuscript is to discuss the results of stress testing studies conducted during the development of denagliptin and the elucidation of its key degradation pathway.

Full text of DOI:10.1002/jps.22069

Elucidating the Pathways of Degradation of Denagliptin
BIREN K. JOSHI,1 BEVERLY RAMSEY,2 BYRON JOHNSON,1 DANIEL E. PATTERSON,3 JEREMIAH POWERS,3
KEVIN L. FACCHINE,1 MARTIN OSTERHOUT,3 MICHAEL P. LEBLANC,1 RENETTA BRYANT-MILLS,2 ROYSTON C.B. COPLEY,4
SCOTT L. SIDES1
1Analytical Sciences, Chemical Development, Five Moore Drive, PO Box 13398, GlaxoSmithKline, RTP, North Carolina 27709
2Product Development, Pharmaceutical Development, Five Moore Drive, PO Box 13398, GlaxoSmithKline, RTP,
North Carolina 27709
3Synthetic Chemistry, Chemical Development, Five Moore Drive, PO Box 13398, GlaxoSmithKline, RTP, North Carolina 27709
4Analytical Chemistry, GlaxoSmithKline, NFSP (N), Third Avenue, Harlow, Essex CM19 5AW, UK
Received 1 October 2009; revised 12 November 2009; accepted 24 November 2009
Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.22069
ABSTRACT: Stress testing or forced degradation studies of denagliptin (1) tosylate in solution
and solid-state, its blends with excipients, and capsules were conducted in order to elucidate
degradation pathways, aid formulation development, and generate data to support regulatory
filings. In solution, denagliptin was stressed in acid, water, and base using organic cosolvents. In
the solid-state, denagliptin was stressed under heat, humidity, and light. Blends of denagliptin
with various excipients were stressed under heat and humidity in order to evaluate whether
tablet was a viable dosage form. Capsules were stressed under heat, humidity, and light. It was
found that denagliptin was stable in the solid-state, but degraded in solution, in blends with all
excipients, and in capsules predominantly by cyclization to (3S,7S,8aS) amidine (2), which
epimerized to (3S,7S,8aR) amidine (3). (3S,7S,8aR) amidine (3) subsequently hydrolyzed to the
corresponding diketopiperazine (4). The purpose of this manuscript is to discuss the results of
stress testing studies conducted during the development of denagliptin and the elucidation of its
key degradation pathway. ß 2010 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci
99:3030–3040, 2010
Keywords: chemical stability; formulation; mass spectrometry; NMR spectroscopy; pre-
formulation; solid state stability; stability
INTRODUCTION
The results of stress testing studies of denagliptin,
preparation of the three predominant degradation
products, and their structure elucidation are dis-
cussed in this manuscript. Intramolecular cyclization
of amino and carboxylic acid, ester, or amide groups
to diketopiperazines has been reported as a key
degradation pathway for model dipeptides,3,4 and
dipeptide drugs like moexipril,5–8 quinapril,9–12
enalapril,13–15 and lisinopril.16 Detailed mechanisms
of diketopiperazine formation by cyclization of di-
peptide amides17 and esters18 has also been reported
in the literature. However, to our knowledge, this is
the first example of drug degradation by intramole-
cular cyclization to amidines.
Denagliptin (Scheme 1), (2S,4S)-4-Fluoro-1-[4-fluoro-
b-(4-fluorophenyl)-L-phenylalanyl]-2-pyrrolidinecar-
bonitrile 4-methylbenzene sulfonic acid salt (1:1), was
being developed for the treatment of type 2 diabetes
due to its ability to inhibit dipeptidyl peptidase IV.
Stress testing studies of denagliptin in solution, in the
solid-state, in blends with various excipients, and
bead-filled capsules were conducted to elucidate its
degradation pathways, aid formulation development,
and generate data for regulatory filings. The condi-
tions used for stress testing and the threshold used for
identification of degradation products were based on
guidance provided in regulatory documents.1,2
Three key degradation products were observed
arising primarily by intramolecular cyclization of
amino and cyano groups in denagliptin (Scheme 1).
EXPERIMENTAL
Materials and Methods
Correspondence to: Biren K. Joshi (Telephone: 919-483-3468;
Fax: 919-483-0443; E-mail: bkj71882@gsk.com)
Denagliptin drug substance and capsules were
manufactured in-house (Glaxo SmithKline, Research
Triangle Park, NC). Monobasic sodium phosphate,
Journal of Pharmaceutical Sciences, Vol. 99, 3030–3040 (2010)
ß 2010 Wiley-Liss, Inc. and the American Pharmacists Association
3030
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 7, JULY 2010

ELUCIDATING THE PATHWAYS OF DEGRADATION OF DENAGLIPTIN
3031
O
O
NH
F
NH
1
H
N
H
N
CN
O
H
8
6
F
F
F
NH
NH
NH
N
F
Hydrolyze
F
8a
4
F
Epimerize
2
N
5
7
3
H2N
O
O
F
F
F
F
F
Denagliptin (1)
(3S,7S,8aS) Amidine (2)
(3S,7S,8aR) Amidine (3)
(3S,7S,8aR) Diketopiperazine (4)
Scheme 1. Pathway of denagliptin degradation.
dibasic sodium phosphate, sodium hydroxide, hydro-
chloric acid, trifluoroacetic acid, dimethyl sulfoxide
(DMSO), 1-methyl 2-pyrrolidinone (NMP), and 2-
methoxy ethyl ether (Diglyme) were obtained from
J.T. Baker (Springfield, NJ). All chemicals were of
reagent grade and used as received. Solvents used for
chromatography were HPLC grade. Deuterated
DMSO was obtained from Cambridge Isotope Labora-
tories (Andover, MA). Mannitol was obtained from
Roquette Pharma (Keokuk, IA), croscarmellose
sodium from FMC Biopolymer (Philadelphia, PA),
crospovidone from International Speciality Products
Inc. (Wayne, NJ), magnesium stearate from Peter
Greven (Bad Munstereifel, Germany), stearic acid
from Mallinckrodt (Hazelwood, MO), sodium stearate
from Witco Corporation (Greenwich, CT), alginic acid
from Edward Mendell (Carmel, NY), and citric
acid from Sigma–Aldrich (St. Louis, MO). Ovens
were obtained from Baxter Scientific products and the
light chamber was obtained from Forma Scientific
Inc. (Model 3890). Amber flint glass vials (part
number 223763) were obtained from Wheaton Scien-
tific Products (Millville, NJ).
flint glass vials. One vial was sealed with rubber
septum and aluminum flange collar whereas the
other was left open. Both vials were placed in a micro-
dessicator containing saturated sodium chloride
solution, which was then placed in the 808C oven.
In order to prepare samples for stressing in light,
60 mg of drug was weighed directly into two clear
polypropylene petri-plates, the drug was spread into a
thin film using spatula, and the plates were placed
uncovered in UV and fluorescent light chambers,
respectively.
Excipient Compatibility Stress Testing Study
The excipient compatibility stress testing study of
denagliptin was conducted using the same drug to
excipient ratios as desired for the 7.5 mg tablet.
Binary, ternary, and quaternary blends of denaglip-
tin were prepared and stressed at 608C and 608C/75%
RH for 2 days in order to evaluate the impact of single
and a combination of excipients on the stability of
denagliptin. Appropriate amounts of drug and
excipient(s) were weighed directly into clear 25-mL
glass vials, the contents were vortexed to mix, and
stored sealed (608C) or open (608C/75% RH) in a
dessicator containing saturated brine solution. The
ratios of drug to various excipients used for prepara-
tion of the blends are summarized in Table 1.
Amounts of sodium stearate and stearic acid were
adjusted so that two equivalents were used relative to
magnesium stearate. Alginic acid and anhydrous
citric acid were evaluated as acid modifiers in
quaternary blends using amounts to ensure that
same number of moles of both were present in the
blends. Excipient free denagliptin was used as
control. Samples were prepared for HPLC analysis
in duplicate by adding 20.0 mL of 1:1 water:acetoni-
trile directly to vial, vortexing to dissolve, filtering
using 1 mm Gelman GF filter, discarding the first few
milliliters transferring to a HPLC vial, and injecting
8 mL for analysis.
Stress Testing Studies of Denagliptin (1) Drug Substance
In solution, denagliptin was stressed at 808C in 0.1 N
HCl, in water under N2, air, and O2 headspaces, and
in 0.01 N NaOH at a concentration of 2.0 mg/mL.
Organic cosolvents were used to enhance the solubi-
lity of denagliptin in the aqueous reagents used for
the study. Dimethyl sulfoxide was used as a cosolvent
in 0.1 N HCl, NMP was used as a cosolvent in water,
and diglyme in 0.01 N NaOH. Different cosolvents
were used depending on the pH of the study in order
to minimize degradation of solvents, and prevent
interaction between cosolvent or its degradation
products and denagliptin. The ratios of organic to
aqueous reagents were 1:4 DMSO:0.1 N HCl, 1:4
NMP:water, and 1:1 diglyme:0.01 N NaOH.
In the solid-state, denagliptin was stressed for
2 weeks at 808C (ambient RH), 2 weeks at 808C/75%
RH, and under UV (400 Wh/m2) and fluorescent light
(2.4 million lux hours). In order to prepare samples for
stressing at 808C and 808C/75% RH, 60 mg of drug
was weighed directly into two different 25-mL amber
Stress Testing Studies of Denagliptin (1) Capsules
Capsules of 7.5 mg, 15.0 mg, and 30.0 mg strengths,
prepared by coating drug on microcrystalline cellu-
lose spheres, and placebo were stressed for 2 weeks at
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 7, JULY 2010

3032
JOSHI ET AL.
Table 1. Denagliptin Excipient Ratios in Blends
Blend Type
Components
Blend Ratio
7.5:183.5
Binary
Denagliptin:mannitol
Denagliptin:croscarmellose sodium
7.5:5.0
Denagliptin:crospovidone
7.5:4.0
Denagliptin:magnesium stearate
7.5:4.0
Ternary
Denagliptin:mannitol:croscarmellose sodium
Denagliptin:mannitol:magnesium stearate
7.5:183.5:5.0
7.5:183.5:4.0
7.5:183.5:5.0:4.0
7.5:183.5:5.0:3.8
7.5:183.5:5.0:4.1
7.5:183.5:4.0:3.8
7.5:183.5:5.0:3.8:20.0
7.5:183.5:5.0:3.8:2.6
Quaternary
Denagliptin:mannitol:croscarmellose sodium:magnesium stearate
Denagliptin:mannitol:croscarmellose sodium:stearic acid
Denagliptin:mannitol:croscarmellose sodium:sodium stearate
Denagliptin:mannitol:crospovidone:stearic acid
Denagliptin:mannitol:croscarmellose sodium:stearic acid:aliginic acid
Denagliptin:mannitol:croscarmellose sodium:stearic acid:citric acid
Acid modifiers
808C (ambient RH), 2 weeks at 808C/75% RH, and
under UV and fluorescent light. Five capsules each of
drug or placebo were placed in 25-mL amber flint
glass vials for stressing at 808C and 808C/75% RH.
The vials were sealed for storage at 808C and left open
for storage at 808C/75% RH. The vials were placed in a
micro-dessicator containing saturated sodium chlor-
ide solution, which was then placed in an 808C oven.
The samples stressed under light were prepared by
placing five capsules each in two clear polypropylene
petri-plates, which were then placed uncovered in UV
and fluorescent light chambers, respectively. Samples
were prepared for HPLC analysis by opening the
capsule shells, transferring the beads and empty
capsule shells to an appropriate volumetric flask,
adding 3:2 water:acetonitrile till the flask was two-
third full, shaking for 40 min, and diluting to mark
with 3:2 water:acetonitrile. The samples were filtered
as described above in excipient compatibility section
before HPLC analysis.
column (150 mm ꢀ 4.6 mm i.d, 3.5 mm particle size)
with 408C column compartment temperature and
215 nm detector wavelength. Gradient started at
0% B and increased to 95% B after 15.0 min with
1.0 mL/min flow rate. Mobile phase A consisted of
0.05% (v/v) trifluoroacetic acid in 90/10 water/
methanol and mobile phase B contained 0.05% (v/v)
trifluoroacetic acid in 10/90 water/methanol. It was
not possible to accurately quantitate the levels of
(3S,7S,8aS) amidine 2 using either of the two
methods because it epimerized to amidine 3.
LC/MS Experiments
Mass spectral data were acquired with a Thermo
Electron LTQ ITMS (San Jose, CA) and an Agilent Ion
Trap SL (Santa Clara, CA) using positive electrospray
ionization (ESI). Typical parameters for operation of
the Finnigan LTQ were as follows: spray voltage,
nominal 4.5 kV; heated capillary, 3258C; sheath gas,
60 psi; auxillary gas, 20 psi; sweep gas, 6 psi; scan
range, 100–2000 Th (or other appropriate range for
the expected mass of the compounds of interest); total
‘‘microscans,’’ 1; maximum injection time, 10 ms and
ion polarity, positive. The Agilent Ion Trap was
operated with a spray voltage of 3.5 kV, a nebulization
pressure of 70 psi, drying gas temperature of 3508C,
and drying gas flow of 12 L/min. An accumulation
time of 300 ms over a scan range of 100–1000 Th was
used for the ion trap mass analyzer.
HPLC Separations
An Agilent 1100 analytical HPLC system was used for
analyses of stressed drug substance, blends, and
capsule samples. Stressed drug substance samples
were analyzed using Zorbax bonus-RP C18 column
(150 mm ꢀ 4.6 mm i.d., 3.5 mm particle size) with 408C
column compartment temperature and 215 nm detec-
tor wavelength. Mobile phase A consisted of 80/20
20 mM phosphate buffer/acetonitrile at pH 6.1 and
mobile phase B contained acetonitrile. Gradient
started at 0% B and increased to 50% B after
18.0 min with 1.5 mL/min flow rate. This method
was modified to make it LC/MS compatible by using
80/20 10 mM ammonium acetate/acetonitrile at
pH 6.1 instead of phosphate buffered mobile phase
as mobile phase A. HPLC column, gradient, flow rate,
and all other parameters were exactly the same for
the two methods. Stressed blends and capsule
samples were analyzed using Zorbax SB-phenyl
High Resolution Mass Spectrometry Experiments
High resolution mass spectrometry (HRMS) analyses
were performed using an LTQ-Orbitrap Discovery XL
hybrid mass spectrometer (ThermoFisher, San Jose,
CA). HRMS experiments were performed by pneu-
matically assisted electrospray ionization (ESI) in
positive ion detection mode with a nominal Orbitrap
resolving power setting of 30,000 (at m/z 400). The
instrument was operated at a spray voltage of 4.5 kV,
a capillary temperature of 3258C, a capillary voltage
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 7, JULY 2010
DOI 10.1002/jps

ELUCIDATING THE PATHWAYS OF DEGRADATION OF DENAGLIPTIN
3033
of 22 V and a tube lens voltage of 70 V. Gas flow
rates, electronics settings and voltages for the ion
optics were optimized and confirmed prior to per-
forming HRMS experiments. Data acquisition for
full scan experiments was performed in profile mode.
Data acquisition for MS/MS fragmentation experi-
ments using higher energy collisionally activated
dissociation (HCD) were performed in centroid mode
using collision energy settings optimized for each
analyte.
Sample introduction was performed using an
Agilent (Santa Clara, CA) 1200 HPLC/DAD system
employing a linear, reverse-phase gradient chroma-
tographic method with mobile phases comprised of
water, acetonitrile and trifluoroacetic acid. The
1.0 mL/min flow of eluent from the HPLC system
was split to achieve approximately 350 mL/min flow
into the mass spectrometer with the remainder
diverted to waste.
between À0.25 and 0.35 eAÀ3. The absolute structure
˚
parameter refined to 0.02(16).
Preparation of (3S, 7S, 8aS) Amidine (2)
(3S,7S,8aS) amidine (2) was prepared by heating a
2.5 mg/mL solution of denagliptin free base (25.98 mg)
in CHCl3 and FeCl3 (12.49 mg; mole ratio 1:1.2) at
808C for 2 h by modification of a procedure described
in the literature19 to prevent epimerization. A
complex of amidine 2 with FeCl3 precipitated out of
solution and was 98.7% pure after filtration. Amidine
2 with purity of 97.5% was obtained by redissolving
the complex in CHCl3 followed by washing with 4
volumes of 0.01 N NaOH and rotary evaporation of
the organic layer.
Preparation of (3S,7S,8aR) Amidine (3)
(3S,7S,8aR) amidine (3) was prepared by weighing
101.64 mg denagliptin free base and 50.6 mg K2CO3
(2 equivalents) directly into a 10-mL reactivial and
adding 10 mL methanol. The vial was placed in a 408C
oven for 19 h after which it was removed, cooled to
ambient temperature, and methanol was evaporated
on a rotary evaporator. The residue was dissolved in
80 mL ethyl acetate and washed with 50 mL water.
Ethyl acetate layer was dried over MgSO4, filtered,
and ethyl acetate was evaporated on a rotary
evaporator. The residue (93.6 mg) was then dissolved
in 2.5 mL CH3CN and 95.0 mg toluene sulfonic acid
monohydrate (2.0 equivalents) was added and the
solution was stirred for 1 h to get a white precipitate of
3 with a purity of 98.6% by HPLC. This sample was
used for NMR analysis. The sample used for single
crystal X-ray diffraction analysis was prepared using
exactly the same ratio of reagents and procedure as
described above starting with 1.0 g denagliptin free
base.
Single Crystal X-Ray Analysis of 31
All measurements were made at 150(2) K using a
Nonius KappaCCD diffractometer with graphite
˚
monochromated Mo Ka radiation (l ¼ 0.71073 A) from
a normal focus sealed tube source. The crystal was
monoclinic, space group P21 (no. 4), with a unit cell of,
˚
˚
3
˚
a ¼ 13.3888(16) A, b ¼ 6.3797(3) A, c ¼ 15.047(3) A,
˚
b ¼ 92.922(15)8, V ¼ 1283.6(3) A , giving Z ¼ 2,
Dcalc ¼ 1.412 Mg mÀ3 and m(Mo Ka) ¼ 0.187 mmÀ1
.
Data were corrected for Lorentz and polarization
effects and a numerical absorption correction was
applied (transmission ¼ 0.950–0.983). The structure
was solved by direct methods and refined (using data
to 458 in 2u) using full-matrix least-squares proce-
ꢀ
ꢁ2
P
dures: the function
w F02 À Fc2 was minimized in
the Bruker-AXS SHELXTL software package (Ver.
6.10 [NT]). Co-ordinates and anisotropic atomic
displacement parameters were refined for all non-
hydrogen atoms, with some appropriate rigid bond
and similarity atomic displacement restraints.
Hydrogen atoms were included in calculated positions
and were refined using a riding model. Isotropic
atomic displacement parameters for the hydrogen
atoms were used as multiples of Ueq for the attached
atoms. The final refinement of 344 variables con-
verged (D/smax ¼ 0.000) to give crystallographic resi-
duals R1 [2210 reflections with F0 > 4s( F0)] and wR2
(all 3129 data) of 0.0754 and 0.1059, respectively. The
‘‘goodness of fit’’ on F2 was 1.093. A final difference
Fourier synthesis showed residual electron density
Preparation of Diketopiperazine (4)
Diketopiperazine formed with 99.0% purity when
98.1 mg denagliptin free base in a 25-mL amber flint
glass vial was heated at 808C for 2 weeks in the solid-
state. This sample was used for NMR analysis.
NMR Experiments
All NMR spectra were recorded at 258C on a VARIAN
UNITY INOVA spectrometer operating at a 1H
frequency of 499.887 MHz and 13C frequency of
125.709 MHz using a nalorac 3 mm indirect detection
probe. Tetramethyl silane (TMS) was used as an
internal chemical shift standard. The samples were
prepared by dissolving 10–15 mg of compound in
250 mL CDCl3 or DMSO-d6, adding 5 mL TMS, and
transferring to a 3 mm NMR tube.
1CCDC 749088 contains the supplementary crystallographic
data for this article. These data can be obtained free of charge
via www.ccdc.cam.ac.uk/data_request/cif, or by emailing data_re-
quest@ccdc.cam.ac.uk, or by contacting The Cambridge Crystal-
lographic Data Centre, 12, Union Road, Cambridge CB2 1EZ, UK;
fax: þ44-122-336033.
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 7, JULY 2010

3034
JOSHI ET AL.
Amidine 3, eluting with retention time of 7.49 min,
was observed at levels between 0.88% and 5.9% area/
area, amidine 2 eluting with 8.36 min retention time
was present at <0.05% to 1.6% area/area, and
diketopiperazine 4 eluting with 9.87 min retention
time was observed at 1.1–2.7% area/area relative to
denagliptin (Table 2).
Base
110
100
Water
Acid
4
6
8
10
12
14
16
Retention time (min)
Figure 1. Overlaid HPLC chromatograms of denagliptin
samples stressed in solution for 6 h at 808C.
Excipient Compatibility Stress Testing Study
Denagliptin was found to be unstable in the presence
of all excipients, especially on exposure to 75%
RH, and degraded to two predominant degradation
products eluting with retention times of 12.22 and
13.93 min, respectively (Figure 2). Denagliptin eluted
with retention time of 11.77 min. Different HPLC
methods were used for analysis of stressed drug
substance and blend samples, which led to different
elution orders and retention times for the same
degradation products. Different methods were used
for analysis of stressed drug substance and drug
product samples, including the excipient compatibil-
ity samples, because different methods were filed in
the regulatory submissions of denagliptin for analysis
of long-term and accelerated stability samples of drug
substance and product. LC/MS analysis of stressed
blend samples indicated that the molecular weight of
12.22 min peak (amidine 3) was 373 Da and 13.93 min
peak (diketopiperazine 4) was 374 Da. Binary blends
of denagliptin with mannitol were unstable and
showed higher degradation than binary blends of
denagliptin with croscarmellose sodium and magne-
sium stearate. All ternary and quaternary blends
containing denagliptin and mannitol blended with
other excipients showed higher degradation than
binary blends prepared by mixing denagliptin and
mannitol. Alginic and citric acid blends also showed
higher degradation than denagliptin drug substance
used as a control. The results, expressed as percent
drop in denagliptin content, are summarized in
Table 3.
RESULTS
Stress Testing Studies of Drug Substance
Denagliptin was stable on stressing in the solid-state,
but unstable on stressing in solution. Three major
impurities were observed in acid and water, and two
major impurities were observed in base, as shown in
Figure 1. LC/MS analysis of stressed samples using
the modified ammonium acetate method indicated
that peaks eluting with retention times 7.49 min
(amidine 3) and 8.36 min (amidine 2) were isobaric to
denagliptin with a molecular weight 373 Da, and the
peak eluting with retention time of 9.87 min had a
molecular weight of 374 Da (diketopiperazine 4).
Accurate mass HRMS measurements of amidine 2
demonstrated a measured m/z value of 374.1477 for
the [M þ H]þ ion to be within 0.5 ppm of calculated
(374.1475) at a resolving power of 34,700 and
indicated a molecular formula of C20H18F3N3O.
Accurate mass measurements of amidine 3 demon-
strated a measured m/z value of 374.1475 for the
[M þ H]þ ion to be within 0.1 ppm of calculated
(374.1475) at a resolving power of 34,800 and
indicated a molecular formula of C20H18F3N3O.
Accurate mass measurements of diketopiperazine
4 demonstrated a measured m/z value of 375.1315
for the [M þ H]þ ion to be within 0.1 ppm of
calculated (375.1315) at a resolving power of 34,600
and consistent with
C20H17F3N2O2.
a
molecular formula of
Table 2. Stress Testing Data for Denagliptin (1) in Solution
0.1 N
0.1 N
Natural pH Degradation (4:1,
0.01 N
NaOH:Diglyme
(1:1), Initial
(%w/w)a
0.01 N
HCl:DMSO
(4:1), Initial
(%w/w)a
HCl:DMSO
(4:1), 808C,
6 h (%w/w)a
Water:NMP), 808C, 6 h (%w/w)a
NaOH:Diglyme,
808C, 6 h (1:1)
(%w/w)a
Compound Name
Initial
N2
92.5
Air
90.9
O2
92.5
Denagliptin
Amidine 3
Amidine 2
99.9
ND
<0.05
ND
93.0
4.1
1.6
100.5
<0.05
<0.05
ND
98.8
<0.05
<0.05
ND
96.3
0.88
<0.05
1.9
5.6
0.67
5.5
0.65
5.9
0.66
2.7
Diketopiperazine 4
Other Impuritiesb
Total
1.1
2.4
2.3
1.2 (9)
101.1
2.4 (13)
102.1
1.1 (10)
101.6
1.6 (13)
102.8
1.5 (9)
100.9
1.6 (10)
103.4
1.3 (10)
100.1
1.8 (11)
100.9
ND, not detected.
aData for initial and stressed samples are expressed in terms of %w/w relative to an external standard and impurities are in %area/area relative to
denagliptin.
bOther impurites include sum of total of synthetic impurites and minor degradation products with numbers in brackets representing the total number of these
impurities.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 7, JULY 2010
DOI 10.1002/jps

ELUCIDATING THE PATHWAYS OF DEGRADATION OF DENAGLIPTIN
3035
200
150
200
150
80C/75%RH
80C
10
11
12
13
14
15
16
10.0 10.5 11.0 11.5 12.0 12.5 13.0 13.5 14.0 14.5 15.0 15.5 16.0
Retention time (min)
Retention time (min)
Figure 2. HPLC chromatogram of denagliptin–mannitol
excipient compatibility sample after 2 days at 608C/75% RH.
Figure 3. Overlaid HPLC chromatograms of 7.5 mg
denagliptin capsule samples stressed for 2 weeks at 808C
and 808C/75% RH.
Stress Testing Studies of Drug Product
amidine (3) was facile and occurred at room
temperature in the presence of FeCl3 and ethanol,
which were used for its synthesis. In addition, three
impurities arising out of reaction between ethanol
and the cyano group of denagliptin were observed at
retention times of 12.84, 13.64, and 13.76 min.
Amidines 2 and 3 eluted with retention times
12.44 and 12.33 min, respectively whereas denaglip-
tin eluted with retention time of 11.95 min. In order
to minimize the side reaction with the solvent and
epimerization of amidine 2 to 3, three aprotic
solvents, tetrahydrofuran (THF), dichloromethane
(CH2Cl2), and chloroform (CHCl3) were evaluated at
temperatures of 60 and 808C. Reactions carried out
in THF and CH2Cl2 were slow and incomplete, but
amidine 2 with purity of 97.5% was obtained when
CHCl3 was used for the reaction.
Denagliptin capsules showed degradation to two
major impurities, eluting with retention times of
12.37 and 14.04 min (Figure 3), after stressing for
2 weeks at 808C and 808C/75% RH. No degradation
was observed on exposure to light. LC/MS analysis of
stressed capsule samples indicated that the molecular
weight of 12.37 min peak (amidine 3) was 373 Da and
14.04 min peak (diketopiperazine 4) was 374 Da.
Amidine 3 was observed at 1.0–1.8% area/area
relative to denagliptin in capsules stressed for 2 weeks
at 808C and at 0.24–0.42% area/area in capsules
stressed for 2 weeks at 808C/75% RH. Diketopiper-
azine 4 was observed at 1.6–3.2% area/area after
stressing the capsules for 2 weeks at 808C and at
3.5–7.9% area/area after stressing for 2 weeks at
808C/75% RH. The levels of degradation products
relative to denagliptin were higher in stressed 7.5 mg
capsules compared to 15.0 mg and 30.0 mg capsules.
Different HPLC systems were used for analysis of
stressed blend and capsule samples resulting in
different retention times for the two key impurities,
but their relative retention time with respect to
denagliptin were the same in both blend and capsule
samples.
NMR Experiments
The 1H NMR spectrum of amidine 2, acquired
immediately after preparation of the sample in
CDCl3, indicated the presence of 10% (by mole) of
amidine 3 as an impurity (Figure 4). During the
course of acquiring 13C, HMQC (one bond H–C
correlation), and HMBC (two to three bond H–C
correlation) data almost complete epimerization of 2
to 3 was observed, which was confirmed by both NMR
and LC/MS analyses (Figure 4). After 48 h in the
NMR spectrometer, resonances attributable to 3 grew
Preparation of (3S,7S,8aS) Amidine (2)
The preparation of (3S,7S,8aS) amidine (2) was
challenging because its epimerization to (3S,7S,8aR)
Table 3. Stress Testing Data for Denagliptin Excipient Compatibility Studies
%Degradation,
608C/75% RH
for 2 Days
%Degradation,
608C for 2 Days
Drug
Diluent/Filler
Disintergrant
—
Lubricant
—
Acidifier
—
Denagliptin
Denagliptin
Denagliptin
Denagliptin
Denagliptin
Denagliptin
Denagliptin
Denagliptin
Denagliptin
Denagliptin
Denagliptin
Denagliptin
—
0.23
0.32
0.17
0.22
0.26
0.63
0.55
2.6
1.7
1.7
3.5
2.6
0.33
1.6
0.5
0.63
1.7
5.9
Mannitol
—
Croscarmellose sodium
—
Croscarmellose sodium
—
Croscarmellose sodium
Croscarmellose sodium
Croscarmellose sodium
Croscarmellose sodium
Croscarmellose sodium
Crospovidone
—
—
—
—
—
Magnesium stearate
—
Magnesium stearate
Magnesium stearate
Stearic acid
Sodium stearate
Stearic acid
Stearic acid
Mannitol
Mannitol
Mannitol
Mannitol
Mannitol
Mannitol
Mannitol
Mannitol
—
—
—
6.4
16.6
12.0
2.7
8.9
16.5
Alginic acid
Citric acid
—
Stearic acid
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 7, JULY 2010

3036
JOSHI ET AL.
9 NH
15
12
H
F
1
NH
8
16
14
8a
2
F
Initial
7
N
13
3
5
11
6
4
10
O
18
19
17
22
21
20
F
5 .0
4 .5
4 .0
3 .5
3 .0
2 .5
2 .0
1 .5
C h e m i c a l S h i f t ( p p m )
48 h
5 .0
4 .5
4 .0
3 .5
3 .0
2 .5
2 .0
1 .5
C h e m i c a l S h i f t ( p p m )
1
Figure 4. Expansions of aliphatic region of H NMR spectra of amidine 2—initial and 48 h.
at the expense of corresponding 2 resonances so all
correlations observed in HMQC and HMBC experi-
ments were attributable to 3. Key correlations shown
by H-3 (d 4.917), H-8a (d 2.602), and H-8 (d 1.732) to C-
1 (d 157.10) and by H-3 (d 4.917), H-10 (d 4.646), and H-
6 (d 3.927) to C-4 (d 168.28) confirmed the presence of
6-member cyclic amidine ring. The chemical shift of
H-8a in 2 was significantly downfield (d 3.971)
compared to 3 (d 2.602), which suggested that C-8a
was the likely site of epimerization. The relative
stereochemistry at C-3, C-7, and C-8a positions in 3
was obtained by single crystal X-ray diffraction
analysis (Figure 5), which confirmed that it was the
3S,7S,8aR isomer based on the known configuration
at C-7. Spectral data of diketopiperazine (4) were
similar to the data for amidine 3. Chemical shift
1
assignments were made by analysis of its H, 13C,
DQCOSY, HMQC, and HMBC data. The presence of 6
member diketopiperazine ring was confirmed by
HMBC correlations shown by H-3 (d 4.624–4.573),
H-8a (d 4.201), and H-8 (d 2.063) to C-1 (d 168.89 ppm)
and by H-3 (d 4.624–4.573) and H-6 (d 3.728) to C-4
C8
C7
C8a
C6
Figure 5. View of cation and anion from crystal structure of amidine 3.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 7, JULY 2010
DOI 10.1002/jps

ELUCIDATING THE PATHWAYS OF DEGRADATION OF DENAGLIPTIN
3037
Table 4. 1H NMR Data of 2, 3, and 4 at 258C
Position
dH (ppm)a,b (2)
dH (ppm)a,b (3)
dH (ppm)a,c (4)
2
Exchanged with solvent
Exchanged with solvent
8.471
7.489
7.432
7.147
7.096
18,22
12,16
19,21
13,15
7
3
10
8a
6
7.442
7.184
6.974
6.900
5.162
5.007
4.606
3.971
4.116
3.285
2.449
1.705
7.509
7.095
6.980
6.926
5.143
4.917
4.646
2.602
3.927
3.448
2.224
1.732
5.279
4.624–4.573
4.624–4.573
4.201
3.728
3.406–3.324
2.350
6
8
8
2.063
aReferenced to TMS ¼ 0.000 ppm.
bAcquired in CDCl3.
cAcquired in DMSO-d6.
(d 164.48). As the resonances for H-3 and H-10
overlapped, the relative stereochemistry of 4 could
not be established by NMR spectroscopy. The
stereochemistry of 4 was determined to be the same
as 3 by an additional single crystal X-ray diffraction
analysis. 1H NMR chemical shift assignment data for
2, 3, and 4 are included in Table 4.
diffraction analysis. As one key purpose of conducting
a stress testing study is to develop a HPLC method
capable of monitoring performance of drug substance
and product under accelerated or long-term condi-
tions, the extent of degradation in stress testing
studies was limited to 5–10% of the active.
Stress Testing Study Results for Denagliptin Drug
Substance
DISCUSSION
Denagliptin was stressed in acid, water, and base in
solution and it degraded to three key cyclization
products under all conditions (Table 2). Hydrolysis of
the cyano group and epimerization of the cyano
stereocenter (C-8a) was observed at low levels in
samples stressed in solution, but not in solid-state
stress testing studies of drug substance or blends and
in capsules stored under stress testing, accelerated or
long term conditions. Hydrolysis of the amide linkage,
epimerization at two other stereocenters, sequence
inversion, and oxidation was not observed for
denagliptin. The preference of denagliptin to undergo
cyclization can be attributed to the basicity of the
amino group (pKa ¼ 6.12; calculated from its pH
solubility curve) and the steric bulk of the N-terminal
residue as reported in the literature3 for cyclization of
C-terminal proline containing peptides. The cycliza-
tion rate for C-terminal proline peptides was shown to
depend on the pKa of the amino group, steric bulk of
the N-terminal residue, ability of the peptide bond to
undergo cis–trans isomerization, and conformational
stability of resulting diketopiperazines.3 The rate of
cyclization of denagliptin would be expected to be
significantly different relative to the rates reported
for dipeptide analogs. The presence of a cyano group
in denagliptin instead of a carboxylic acid, ester, or
amide groups present in dipeptide analogs could
significantly impact the rates of cis–trans isomeriza-
tion of the peptide bond. In addition, the mechanism
Regulatory Requirements for Conducting Stress Testing
Studies
Stress testing studies of denagliptin drug substance
and capsules were performed according to recom-
mendations in the regulatory guidance for stress
testing of new drug substances and products.1 The
guidance indicates that the susceptibility of drug to
hydrolysis must be examined across a wide range of
pH values in solution or suspension. In the solid-state
and in drug products, the stress testing study should
use temperatures higher than used for accelerated
testing (408C), humidities of 75% RH or greater, and
photostability testing on at least one batch of drug
substance and product. The guidance also indicates
that degradation products that are not formed under
accelerated or long-term conditions need not be
evaluated. Data from the study are typically also
used to evaluate mass balance and stereochemical
stability. The thresholds for identification and
qualification of impurities in new drug substance
and products indicated in regulatory guidance2
required the identification of all impurities greater
than 0.2% in stressed samples. As amidine 2, amidine
3, and diketopiperazine 4 were the only three
impurities exceeding identification thresholds on
accelerated stability testing, they were prepared
and characterized by NMR and single crystal X-ray
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 7, JULY 2010

3038
JOSHI ET AL.
of cyclization of denagliptin would also be different
compared to dipeptide analogs because the initially
formed intermediate in aminolysis of cyano group is
not tetrahedral. Therefore the comparison between
denagliptin and dipeptide cyclization is restricted to
the pathway, for example, cyclization by aminolysis,
and not mechanism of degradation.
Denagliptin was stable on stressing in the solid-
state due to the rigidity of its crystal lattice
attributable to hydrogen bonds between denagliptin
cation and tosylate anion in the unit cell, which kept
the amino group protonated and trans to the cyano
group. The solid-state stress testing study was
stopped after 2 weeks of storage at 808C and 808C/
75% RH because the kinetic equivalent of 6 months at
408C/75% RH exposure recommended in the regula-
tory guidance1 was achieved. The extrapolation of
storage time from 808C/75% RH to 408C/75% RH was
based on doubling of reaction rate for every 108C rise
in temperature using a conservative estimate of
12.2 kcal/mol for energy of activation, Ea.20
aqueous solution (i.e., 2.6 mg/mL) of denagliptin
tosylate is ca. 4.3, which is below the apparent pHmax
of 5.3. pHmax is defined as the pH at which both the
salt and free base species of a drug are simultaneously
saturated. Moreover, the surface acidity of mannitol
expressed as pHeq is reported to be 3.23.25 These
factors preclude the formation of solid free base by
disproportionation in blends with mannitol.26 It is
more likely that the degradation of denagliptin–
mannitol binary blend at 608C/75% RH occurred in
microsolutions formed by adsorption of water, which
was exacerbated by higher surface interaction
between drug and mannitol due to low drug:excipient
ratio (1:25). This hypothesis is consistent with greater
degradation on exposure to 608C/75% RH compared to
608C, and the observation that degradation in blends
followed the order sodium stearate > magnesium
stearate > croscarmellose sodium. This order is con-
sistent with the basicity of these excipients and the
base catalyzed nature of cyclization of dipeptides.3,17
Higher surface interaction due to low drug excipient
ratio has also been reported as a cause for cyclization
in enalapril.14 Acidic excipients like stearic acid,
alginic acid, and citric acid did not improve the
stability of the blends significantly, which could be
due to their ability to enhance the rate of epimeriza-
tion of amidine 2 (Scheme 2) and/or hydrolysis of
amidine 3 to diketopiperazine 4 (Scheme 1). Higher
degradation for all stearic acid blends could be
attributed to the melting of stearic acid (mp ꢁ 548C),27
which could dissolve the drug and enhance mobility
relative to a solid. Denagliptin exists as anhydrous
form 1 and no evidence for any other form was
obtained in any of the stressed samples by XRD. Since
denagliptin was found to be chemically unstable in
the presence of all excipients, including mannitol, the
development of tablet was discontinued and bead
filled capsules were progressed for further develop-
ment.
Excipient Compatibility Stress Testing Study
Isothermal excipient compatibility stress study of
denagliptin in combination with various excipients
was performed early in development to evaluate the
feasibility of preparing a stable tablet formulation.
Mannitol was evaluated as a diluent in order to avoid
the use of hygroscopic excipients like microcrystalline
cellulose, which accelerated cyclization of quinapril
and moexipril.7,10 However, just after 2 days of
stressing at 608C/75% RH the denagliptin–mannitol
binary blend showed higher degradation compared to
unblended drug used as a control. Denagliptin–
mannitol binary blend was stable after 2 days of
stressing at 608C under ambient RH, but ternary and
quaternary blends showed higher degradation than
the unblended denagliptin. As mannitol is polar and
protic, it would not be expected to enhance trans–cis
isomerization of denagliptin amide bond21,22 to cause
higher degradation in the binary blends. The rate of
amide bond isomerization in denagliptin and its
blends could not be determined by NMR spectro-
scopy23,24 due to low aqueous solubility of denagliptin
(2.6 mg/mL) and close chemical shifts of cis and trans
isomer resonances. The natural pH of a saturated
Stress Testing Results for Denagliptin Capsules
Amidine 3 and diketopiperazine 4 were key degrada-
tion products observed in stressed capsule samples,
confirming that the key degradation pathway in
capsules was the same as in solution and excipient
blends. However, the rate of degradation was much
NH+2
NH2+
NH2
NH
H
H
N
F
F
F
NH
NH
+
F
H O
H2O
F
F
H3O
+
2
+
N
+
N
O
O
O
F
F
F
Scheme 2. Proposed mechanism for epimerization of amidine 2 in acid.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 7, JULY 2010
DOI 10.1002/jps

ELUCIDATING THE PATHWAYS OF DEGRADATION OF DENAGLIPTIN
3039
slower compared to excipient blends, justifying the
progression of bead filled capsules as the formulated
product.
dose products of moisture sensitive drugs that require
high excipient to drug ratio.
Epimerization of (3S,7S,8aS) Amidine (2)
ACKNOWLEDGMENTS
Results of solution phase stress testing experiments
suggest that epimerization of amidine 2 occurred in
all the samples. The absence of epimerized denaglip-
tin in stressed samples, including samples stressed
for 4 h at 808C and 22 h at 608C (data not reported)
where the extent of denagliptin degradation was
<5%, suggests epimerization occurred predominantly
after cyclization to amidine 2. This observation is
consistent with results reported in the literature for
epimerization of diketopiperazines.28 The synthe-
sized sample of amidine 2 also epimerized rapidly in
CDCl3 in the NMR tube, which supports epimeriza-
tion after cyclization. The rapid epimerization of
amidine 2 can be explained by the possibility of
existence of tautomeric forms in equilibrium (Scheme
2) and steric factors. Cyclic dipeptides containing an
aromatic amino acid residue have been known to
adopt folded F conformation in solution as well as in
the solid-state.29 In this conformation, the aromatic
side-chain ring stacks over the diketopiperazine ring
and is favored by 13 kJ/mol over other possibilities in
polar solutions. If amidine 2 were to adopt this
conformation, the steric interaction between methy-
lene group on 5-member ring and phenyl rings would
be higher due to cis orientation compared to amidine
3, which has hydrogen cis to phenyl ring. The
preference for cis–trans epimerization at ring junc-
tion stereocenter over the ring stereocenter was also
reported for diketopiperazine derivatives of enalapril
and lisinopril.30 Single crystal X-ray diffraction data
for amidine 3 and diketopiperazine 4 are consistent
with aromatic ring stacked over the amidine and
diketopiperazine rings, respectively in the solid-state.
It may be possible for them to adopt a similar
conformation in CDCl3 solution.31
Professor William Clegg and Dr. Sophie Dale
(University of Newcastle) are thanked for the
GlaxoSmithKline funded X-ray crystallographic
analysis of Amidine 3. Dr. Richard Winnike (Glaxo-
SmithKline) and Dr. Anjali Joshi (Eisai Inc.) are
thanked for helpful discussions.
REFERENCES
1. Q1A (R2) Stability testing of new drug substances
and products, November 2003. http://www.fda.gov/downloads/
Drugs/GuidanceComplianceRegulatoryInformation/Guidances/
ucm073369.pdf.
2. Q3B (R) Impurities in new drug products (revision 2),
August 2006. http://www.fda.gov/downloads/Drugs/Guidance-
ComplianceRegulatoryInformation/Guidances/ucm073389.pdf.
3. Goolcharan C, Borchardt RT. 1998. Kinetics of diketopipera-
zine formation using model peptides. J Pharm Sci 87:283–287.
4. Leung SS, Grant DJW. 1997. Solid state stability studies of
model dipeptides: Aspartame and aspartylphenylalanine.
J Pharm Sci 86:64–71.
5. Gu L, Strickley RG. 1987. Diketopiperazine formation, hydro-
lysis, and epimerization of the new dipeptide angiotensin-
converting enzyme inhibitor RS-10085. Pharm Res 4:392–397.
6. Strickley RG, Visor GC, Lin L, Gu L. 1989. An unexpected pH
effect on the stability of moexipril lyophilized powder. Pharm
Res 6:971–975.
7. Gu L, Strickley RG, Chi L, Chowan ZT. 1990. Drug-excipient
incompatibility studies of the dipeptide angiotensin-converting
enzyme inhibitor, moexipril hydrochloride: Dry powder vs wet
granulation. Pharm Res 7:379–383.
8. Gu L, Strickley RG. 1990. A profound solvent effect on the
diketopiperazine formation of the new dipeptide angiotensin-
converting enzyme inhibitor, moexipril. Int J Pharm 60:99–
107.
9. Guo Y, Byrn SR, Zografi G. 2000. Physical characteristics and
chemical degradation of amorphous quinapril hydrochloride.
J Pharm Sci 89:128–143.
10. Guo Y, Byrn SR, Zografi G. 2000. Effects of lyophilization on the
physical characteristics and chemical stability of amorphous
quinapril hydrochloride. Pharm Res 17:930–935.
11. Li J, Guo Y, Zografi G. 2002. The solid-state stability of amor-
phous quinapril in the presence of b-cyclodextrins. J Pharm Sci
91:229–243.
12. Li J, Guo Y, Zografi G. 2002. Effects of a citrate buffer system on
the solid-state chemical stability of lyophilized quinapril pre-
parations. Pharm Res 19:20–26.
13. Wang S, Lin S, Chen T. 2001. Reaction kinetics of solid-state
cyclization of enalapril maleate investigated by isothermal
FT-IR microscopic system. Chem Pharm Bull 49:402–406.
14. Cotton ML, Wu DW, Vadas EB. 1987. Drug-excipient interac-
tion study of enalapril maleate using thermal analysis and
scanning electron microscopy. Int J Pharm 40:129–142.
15. Wang S, Lin S, Chen T, Cheng W. 2004. Eudragit E accelerated
the diketopiperazine formation of enalapril maleate deter-
mined by thermal FTIR microspectroscopic technique. Pharm
Res 21:2127–2132.
Summary
Denagliptin degraded by cyclization on stressing in
solution, excipient blends, and bead filled capsules.
The structures of three key degradation products
were elucidated by analysis of NMR and single crystal
X-ray diffraction data. Amidine 2 was unstable and
epimerized to amidine 3, which hydrolyzed further to
the corresponding diketopiperazine 4. All the blends
of denagliptin with excipients, including acidic and
basic excipients, showed higher degradation com-
pared to unblended denagliptin. Bead filled capsules
of denagliptin were developed in order to minimize
the interaction of denagliptin with excipients and
moisture. The capsules showed adequate stability
and may be a good alternative for formulating low
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 7, JULY 2010

3040
JOSHI ET AL.
16. Wang S, Lin S, Chen T. 2000. Thermal dependant dehydration
process and intramolecular cyclization of lisinopril dihydrate in
the solid state. Chem Pharm Bull 48:1890–1893.
17. Capasso S, Vergara A, Mazzarella L. 1998. Mechanism of
2,5-dioxopiperazine formation. J Am Chem Soc 120:1990–
1995.
25. Scheef CA, Oelkrug D, Schmidt PC. 1995. Surface acidity of
solid pharmaceutical excipients. III. Excipients for solid dosage
forms. Eur J Pharm Biopharm 46:209–213.
26. Guerrieri P, Taylor LS. 2009. Role of salt and excipient proper-
ties on disproportionation in the solid-state. Pharm Res
26:2015–2026.
18. Purdie JE, Benoiton NL. 1973. Piperazinedione formation
from esters of dipeptides containing glycine, alanine, and
sarcosine: The kinetics in aqueous solution. J Chem Soc II
1845–1852.
27. Wade A, Weller PJ. 1994. Handbook of pharmaceutical exci-
pients. 2nd edition. Washington DC: American Pharmaceutical
Association and. Royal Pharmaceutical Society of Great
Britain. pp. 494–497.
19. Fuks R. 1973. N-Alkylation of nitriles: I. Tetrahedron 29:2147–
2151.
20. Connors KA, Amidon GL, Stella VJ. 1986. Chemical stability of
pharmaceuticals. 2nd edition. New York: John Wiley & Sons.
p. 24.
21. Eberhardt ES, Loh SN, Hinck AP, Raines RT. 1992. Solvent
effects on the energetics of prolyl peptide bond isomerization.
J Am Chem Soc 114:5437–5439.
22. Fischer G. 2000. Chemical aspects of peptide bond isomerisa-
tion. Chem Soc Rev 29:119–127.
23. Grathwohl C, Wuthrich K. 1981. NMR studies of the rates of
proline cis-trans isomerization in oligopeptides. Biopolymers
20:2623–2633.
24. Sui Q, Borchardt D, Rabenstein DL. 2007. Kinetics and equili-
bria of cis/trans isomerization of backbone amide bonds in
peptoids. J Am Chem Soc 129:12042–12048.
28. Gaines SM, Bada JL. 1988. Aspartame decomposition and
epimerization in the diketopiperazine and dipeptide products
as a function of pH and temperature. J Org Chem 53:2757–
2764.
29. Liwo R, Ciarkowski J. 1985. Origin of the ring-ring interaction
in cyclic dipeptides incorporating an aromatic amino acid. Tett
Lett 26:1873–1876.
30. Demeter A, Fodor T, Fischer J. 1998. Stereochemical
investigations on the diketopiperazine derivatives of enalapril
and lisinopril by NMR spectroscopy. J Mol Struct 471:161–
174.
31. Suaifan GA, Mahon MF, Arafat T, Threadgill MD. 2006. Effects
of steric bulk and stereochemistry on the rates of diketopiper-
azine formation from N-aminoacyl-2,2-dimethylthiazolidine-4-
carboxamides (Dmt dipeptide amides)—A model for a new
prodrug linker system. Tetrahedron 62:11245–11266.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 7, JULY 2010
DOI 10.1002/jps