The Hydrolysis of Diclofenac Esters: Synthetic Prodrug Building Blocks for Biodegradable Drug-Polymer Conjugates

Article abstract of DOI:10.1002/jps.24665

Degradation reactions on diclofenac-monoglycerides (3a,b), diclofenac-(p-hydroxybenzoate)-2-monoglyceride (3c), diclofenac (1), and diclofenac lactam (4) were performed at 37°C in isotonic buffer solutions (apparent pH range 1-8) containing varying concentrations of acetonitrile (ACN). The concentration remaining of each analyte was measured versus time. Diclofenac-monoglycerides and diclofenac-(p-hydroxybenzoate)-2-monoglyceride (3c) were both found to undergo facile and complete hydrolysis in pH 7.4 isotonic phosphate buffer/10% ACN. Under mildly acidic, neutral or alkaline conditions, diclofenac-(p-hydroxybenzoate)-2-monoglyceride (3c) had the fastest hydrolysis rate (t_1/2 = 3.23 h at pH 7.4), with simultaneous formation of diclofenac lactam (4) and diclofenac (1). Diclofenac-monoglycerides (3a,b) hydrolyzed more slowly under the same conditions, to again yield both diclofenac (1) and diclofenac lactam (4). There was also transesterification of diclofenac-2-monoglyceride (3b) to its regioisomer, diclofenac-1-monoglyceride (3a) across the pH range. Diclofenac was shown to be stable in neutral or alkaline conditions but cyclized to form the lactam (4) in acidic conditions. Conversely, the lactam (4) was stable under acidic conditions but was converted to an unknown species under alkaline or neutral conditions.

Full text of DOI:10.1002/jps.24665

Journal of Pharmaceutical Sciences 105 (2016) 773e785
Contents lists available at ScienceDirect
Journal of Pharmaceutical Sciences
journal homepage: www.jpharmsci.org
Pharmaceutics, Drug Delivery and Pharmaceutical Technology
The Hydrolysis of Diclofenac Esters: Synthetic Prodrug Building Blocks
for Biodegradable DrugePolymer Conjugates
Feng Wang 1, Joshua Finnin 1, Cassandra Tait 1, Stephen Quirk 2, Igor Chekhtman 1,
,
,
,
, *
Andrew C. Donohue 1 3, Sarah Ng 1 3, Asha D'Souza 1 3, Russell Tait 3, Richard Prankerd 1
1 Monash Institute of Pharmaceutical Sciences, Monash University (Parkville Campus), Parkville, Victoria 3052, Australia
2 Kimberly-Clark Corporation, Roswell, Georgia 30022
3 PolyActiva P/L, Melbourne, Victoria 3000, Australia
a r t i c l e i n f o
a b s t r a c t
Article history:
Degradation reactions on diclofenac-monoglycerides (3a,b), diclofenac-(p-hydroxybenzoate)-2-
monoglyceride (3c), diclofenac (1), and diclofenac lactam (4) were performed at 37ꢀC in isotonic
buffer solutions (apparent pH range 1-8) containing varying concentrations of acetonitrile (ACN). The
concentration remaining of each analyte was measured versus time. Diclofenac-monoglycerides and
diclofenac-(p-hydroxybenzoate)-2-monoglyceride (3c) were both found to undergo facile and complete
hydrolysis in pH 7.4 isotonic phosphate buffer/10% ACN. Under mildly acidic, neutral or alkaline condi-
tions, diclofenac-(p-hydroxybenzoate)-2-monoglyceride (3c) had the fastest hydrolysis rate (t1/2 ¼ 3.23 h
at pH 7.4), with simultaneous formation of diclofenac lactam (4) and diclofenac (1). Diclofenac-
monoglycerides (3a,b) hydrolyzed more slowly under the same conditions, to again yield both diclofe-
nac (1) and diclofenac lactam (4). There was also transesterification of diclofenac-2-monoglyceride (3b)
to its regioisomer, diclofenac-1-monoglyceride (3a) across the pH range. Diclofenac was shown to be
stable in neutral or alkaline conditions but cyclized to form the lactam (4) in acidic conditions.
Conversely, the lactam (4) was stable under acidic conditions but was converted to an unknown species
under alkaline or neutral conditions.
Received 15 December 2014
Revised 1 September 2015
Accepted 8 September 2015
Available online 5 November 2015
Keywords:
hydrolysis
pH rate profile
diclofenac esters
wound care
pendent drug attachment
prodrugs
controlled release
kinetics
stability
© 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
polymeric drug delivery systems
Introduction
The relative rates of hydrolysis (to liberate the parent drug) are
anticipatedtobesignificantlydifferent, asthearylesterisanactivated
This article reports the synthesis, hydrolytic behavior, and prod-
uct formation for two diclofenac esters, diclofenac-monoglyceride
[as an equilibrated mixture of two regioisomers DCF-1-MG, 3a,
and DCF-2-MG, 3b (throughout this manuscript, the mixture of
regioisomers of DCF-MG will be referred to as DCF-MG (3a,b),
whereas DCF-1-MG (3a) or DCF-2-MG (3b) will be used when
referring to a specific individual regioisomer.)] and diclofenac-
(p-hydroxybenzoate)-2-monoglyceride (DCF-PHB-MG, 3c) as typical
monomeric building blocks for the synthesis of drugepolymer
conjugates (DPCs).1 Two products arose from the hydrolysis of both
thealkylandarylesters, diclofenac(DCF,1)and diclofenaclac-tam(4).
ester with a good leaving group. It was hoped that from an under-
standing of the monomer hydrolysis kinetics the rates of drug release
from a DPCs could be tailored. The structure of DCFand the DCF esters
suggested the possibility of lactam formation by an intramolecular
condensation (dehydration) reaction. Despite this possibility, as far as
we are aware, there have been no previous reports of lactam forma-
tion directly from the hydrolysis of esters of DCF,2 though there have
been reports of lactam formation during synthesis of diclofenac an-
alogues, particularly activated esters of diclofenac. Along with diclo-
fenac (1) and diclofenac lactam (4), the ester's stability and
degradation/hydrolytic behavior were studied in solutions of isotonic
buffers with apparent pH values in the range 1-8, using sufficient
acetonitrile (ACN) cosolvent to give adequate solubility for uncom-
plicated analysis.
This article contains supplementary material available from the authors upon
request or via the Internet at http://dx.doi.org/10.1002/jps.24665.
* Correspondence to: Richard Prankerd (Telephone: þ61-3-99039520; Fax: þ61-
3-96570777).
Diclofenac is a potent nonsteroidal anti-inflammatory drug
(NSAID) that is widely used for the treatment of pain and inflam-
mation.3,4 In addition to this primary activity, studies have shown
that DCF is a broad spectrum antimicrobial agent.5,6 NSAIDs like
E-mail address: richard.prankerd@monash.edu (R. Prankerd).
http://dx.doi.org/10.1002/jps.24665
0022-3549/© 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.

774
F. Wang et al. / Journal of Pharmaceutical Sciences 105 (2016) 773e785
DCF are also known to exert an antiproliferative effect on fibroblasts
and keratinocytes,7 which may counteract their anti-inflammatory
effect in the early stages of the wound healing cycle. Further studies
in acute wounds have demonstrated that even though DCF reduced
fibroblast numbers in an incisional acute wound, DCF did not
impair normal healing.8 The observation that DCF has potential
therapeutic activities important to the management of chronic
wounds led to the intriguing possibility that it would be clinically
efficacious to deliver therapeutic levels of DCF into chronic
wounds.9
Experimental
Materials
Diclofenac was purchased from Beta Pharm Company Ltd.
(batch number: 201006280; assay: 99.12%; Shanghai, China). All
other solvents or reagents were HPLC or analytical reagent grade.
Supplementary Information contains further information for HPLC
method validation, preparation of lactam (4), and additional kinetic
data.
We sought to build DPCs where DCF was attached pendent to
the polymer backbone through an ester linkage, which is labile in
Synthesis of Diclofenac Esters
a
physiological environment and provides a mechanism for
controlled release. The polymer could be processed into fiber form
and incorporated into wound care dressings with a key perfor-
mance criterion that they must release therapeutic amounts of DCF
in a controlled manner when exposed to wound exudate.
Synthesis of the diclofenacepolymer conjugate involves reac-
tion of diclofenac conjugated via an ester linkage to a diol, as a drug
incorporating monomer, with a comonomer (i.e., a diisocyanate) to
Preparation of 1,3-Dihydroxypropan-2-yl 2-(2-((2,6-Dichlorophenyl)
Amino)Phenyl)Acetate (DCF-2-MG)(3b)
To a mixture of diclofenac (11.47 g, 38.7 mmol), DMAP (0.25, 2.0
mmol), and cis-1,3-benzylidene glycerol (6.99 g, 38.8 mmol), a so-
lution of DCC (10.01 g, 48.5 mmol) in anhydrous DCM (500 mL) was
added drop wise at 0ꢀC over a period of 30 min. The reaction was
stirred at 0ꢀC for 2 h. The crude material was purified via column
chromatography on silica gel (20% ethyl acetate/hexanes as eluent)
to give 2-phenyl-1,3-dioxan-5-yl 2-{2-[(2,6-dichlorophenyl)amino]
phenyl}acetate in 68% yield as an off-white solid. 1H NMR (200 MHz,
CDCl3) * 7.71-7.27 (m, 8H), 7.27-7.09 (m, 1H), 7.09-6.88 (m, 3H), 6.63
(d, J ¼ 7.9 Hz, 1H), 5.59 (s, 1H), 4.78 (s, 1H), 4.26 (dd, J ¼ 40.1, 12.6 Hz,
4H), 4.01 (s, 2H). 13C NMR (50 MHz, CDCl3) * 172.65, 142.88, 137.91,
137.86, 131.09, 129.52, 129.12, 128.90, 128.33, 128.14, 126.10, 124.42,
124.07, 122.19, 118.45, 101.17, 68.92, 66.91, 38.60. ESI-MS: m/z 460
produce
a polyurethane-based DPC. The drugincorporating
monomer is a prodrug, the hydrolytic behavior of which can be
studied independently to confirm the lability of the ester linkage
and to better understand diclofenac release from the final DPC
product.
The purpose of this study was to (1) confirm that hydrolysis of
an aryl ester construct was faster than an alkyl ester, (2) show that
DCF was the predominant product of hydrolysis, (3) examine the
stability of DCF and its lactam alone, (4) understand the relation-
ships between rates of hydrolysis and pH, and (5) explore the utility
of the constructs for design of a wound-responsive product.
Preparation of DPCs of DCF has previously been attempted. Two
groups have described the use of standard ester linkages with
nonbiodegradable polymethacrylate polymer backbones.10,11 In
both cases, the drug release profile was measured with a direct UV
method that would not differentiate between DCF and its lactam
degradation product.
A number of other diclofenac alkyl ester prodrugs with ester
linkages have been made.10-15 Bonina et al.13 reported chemical
hydrolysis half-lives in the range 400e500 h (pH 7.4 buffer, 32ꢀC)
for an homologs series of polyoxyethylene ester prodrugs. Tam-
mara et al.15 produced a series of morpholinoalkyl esters of DCF
with reported chemical hydrolysis half-lives in the range 3e34 h
(pH 7.4 buffer, 37ꢀC). Both the disappearance of the prodrug and the
formation of DCF were measured, with mass balance achieved.
Jilani et al.14 made some hydroxyethyl esters of DCF and reported a
chemical hydrolysis half-life of about 36 h (pH 7.4, 37ꢀC). Bonina
et al.,13 Tammara et al.,15 and Jilani et al.14 all used HPLC methods to
quantify the prodrug separately from DCF and none reported any
observations regarding the formation of lactam (4).
37
37
(3%, Mþ, C24
H H
Cl2NO4), 459 (13%, Mþ, C24 Cl35ClNO4), 457 (15%,
21 21
35
Mþ, C24
H
21
Cl2NO4), 242 (11), 214 (100), 103 (15). IR nmax (cmꢁ1):
3320, 2855, 1717, 1504, 1451, 1142, 1080, 908, 728, 697.
1-(2,6-Dichlorophenyl)indolin-2-one (diclofenac lactam) (4)
was also isolated as a by-product in 30% yield. Spectral data were
consistent both with that reported and an authentic sample (see
Supporting Information).
2-Phenyl-1,3-dioxan-5-yl
2-{2-[(2,6-dichlorophenyl)amino]
phenyl}acetate (3.13 g, 6.8 mmol), 10% (w/w) palladium on carbon
(0.31 g) in ethyl acetate (60 mL) was hydrogenated under one at-
mosphere of hydrogen (balloon) for 16 h at room temperature. The
catalyst was removed by filtration through Celite. The crude ma-
terial was purified by reslurrying and filtration from 30% ethyl ac-
etate/hexanes. The title compound (3b) was obtained in 71% yield
as an off-white solid. 1H NMR (200 MHz, DMSO) * 7.68-7.42 (m, 2H),
7.29-6.91 (m, 4H), 6.91-6.69 (m, 1H), 6.25 (d, J ¼ 7.8 Hz, 1H), 4.90-
4.62 (m, 3H), 3.79 (s, 2H), 3.65-3.39 (m, 4H). 13C NMR (50 MHz,
DMSO) * 171.93, 143.27, 137.57, 131.29, 131.21, 131.05, 129.58, 128.12,
126.32, 123.89, 121.12, 116.33, 76.87, 60.15, 37.72. ESI-MS: m/z 373
37
37
(2%, Mþ, C17
H H
Cl2NO4), 371 (12%, Mþ, C17 Cl35ClNO4), 369 (18%,
17 17
35
Mþ, C20
H Cl2NO4), 295 (4), 279 (11), 277, (16), 242 (29), 216 (37)
21
214 (100), 180 (13). IR nmax (cmꢁ1): 3285, 2943, 1708, 1579, 1509,
Other NSAIDepolymer conjugates16 have been made to com-
pare hydrolysis rates for alkyl esters with aryl esters, with faster
drug release observed with the aryl ester. Also, some evidence
suggesting release rates were greater in more strongly basic media
was found. This has important implications for wound care, as it
offers the potential to develop a wound-responsive product.17 The
pH of normal skin is usually approximately 5.5, with a normal range
of 4.0-7.018 and the pH of serum (or acute wound exudate) is 7.35-
7.4519; chronic wound exudate is marked by an elevated pH, often
1450, 1289, 1046, 770, 743.
Preparation of 1,3-Dihydroxypropan-2-yl 4-(2-(2-((2,6-Dichlorophenyl)
Amino)Phenyl)Acetoxy) Benzoate (3c)
Using the procedure described above, 1,3-dihydroxypropan-2-yl
4-(2-(2-((2,6-dichlorophenyl)amino)phenyl)acetoxy) benzoate (3c)
was prepared in two steps from diclofenac in 62% and 67%, respec-
tively. Diclofenac lactam (4) was also isolated as a by-product from
the first step in 37% yield. 1H NMR (200 MHz, CDCl3) * 8.26-7.86 (m,
2H), 7.34 (d, J ¼ 8.0 Hz, 3H), 7.23-7.09 (m, 3H), 7.07-6.88 (m, 2H),
6.70-6.48 (m, 2H), 5.14 (p, J ¼ 4.7 Hz, 1H), 4.07 (s, 2H), 3.94 (d, J ¼ 4.7
Hz, 4H), 2.50 (bs, 2H). 13C NMR (50 MHz, CDCl3) * 170.10, 154.43,
142.66, 137.63, 131.38, 131.03, 129.44, 128.85, 128.41, 127.54, 124.21,
123.48, 122.37, 121.71, 118.59, 75.81, 62.43, 38.53. ESI-MS: m/z 493
in the range 8.0-8.9.20,21
A wound bed microenvironment-
responsive product would release more drug during the periods
of active wound exudation, driven by the high pH of the wound,
and less drug as pH of the wound microenvironment decreases
because of wound healing. That is, ideally the DCF release rate
profile is pH dependent.
37
37
(2%, Mþ, C24
H H
Cl2NO6), 491 (6%, Mþ, C24 Cl35ClNO6), 489 (8%, Mþ,
21 21

F. Wang et al. / Journal of Pharmaceutical Sciences 105 (2016) 773e785
775
35
C24
H
Cl2NO6), 280 (22), 279 (36), 278 (34), 277 (49), 250 (20), 216
DCF-2-monoglycerides, and subsequently, the rates of trans-
esterification between the two structural isomers. The molar ratio
of isomers present in the starting material differed between ex-
periments. Isomer ratios of approximately 10:90 and 90:10 (3a:3b)
were used for experiment 1 and 2, respectively. The concentrations
of 3a and 3b were measured as a function of time using the HPLC
procedure described below. Nonlinear least-squares regression
(weighting factor 2) was performed on the experimental data using
Micromath Scientist® 3.0 software (St. Louis, Missouri). First-order
rate constants describing transesterification and hydrolysis were
estimated from this regression.
21
(40), 214 (100),121 (83). IR nmax (cmꢁ1): 3541, 3326, 2981, 1741,1703,
1504, 1453, 1302, 1230, 1121, 750.
Preparation of 2,3-Dihydroxypropyl 2-(2-((2,6-Dichlorophenyl)
Amino)Phenyl)Acetate (DCF-1-MG) (3a)
The general procedure for DCC coupling outlined above was
used. Reaction of diclofenac and solketal gave (2,2-dimethyl-1,3-
dioxolan-4-yl)methyl 2-(2-((2,6-dichlorophenyl)amino)phenyl)ac-
etate in 74% yield as a clear yellow oil. Diclofenac lactam (4) was
also isolated as a by-product in 21% yield.
(2,2-Dimethyl-1,3-dioxolan-4-yl)methyl
2-(2-((2,6-dichloro-
phenyl)amino)phenyl) acetate (3.47 g, 8.4 mmol) was stirred in 80%
aqueous acetic acid at 80ꢀC for 5 h. The crude mixture was purified
via column chromatography on silica gel using 50% ethyl acetate/
hexane as eluent. The title compound, 2,3-dihydroxypropyl 2-(2-
((2,6-dichlorophenyl)amino)phenyl)acetate (DCF-1-MG) (3a) was
obtained in 71% yield as a clear yellow oil. 1H NMR (200 MHz,
CDCl3) * 7.43-7.03 (m, 4H), 7.03-6.86 (m, 2H), 6.79 (bs, 1H), 6.64-
6.45 (m, 1H), 4.32-4.11 (m, 2H), 4.03-3.71 (m, 3H), 3.71-3.41 (m,
2H), 3.36 (s, 2H). 13C NMR (50 MHz, CDCl3) * 172.76, 142.68, 137.70,
Analytical Procedure
A Waters® HPLC 2690 gradient separation module was equip-
ped with a Waters Symmetry® C18 column (15 ꢂ 0.46 cm2, 5
mm
particle size) and a Waters® 2487 dual wavelength UVeVis de-
tector (lanal ¼ 275 and 245 nm). The mobile phase comprised
either ACNewatereacetic acid (47:52:1, v/v) or ACNewatereacetic
acid (55:45:1, v/v). The latter mobile phase was used for analysis of
DCF-PHB-MG (3c), to achieve a shorter elution time. Two sets of
standard solution mixtures, each monomer mixed with diclofenac
(1) and lactam (4), were prepared in the mobile phase. These so-
lutions were suitably diluted with the mobile phase to give a
130.88, 129.55, 128.87, 128.13, 124.18, 123.99, 122.07, 118.27, 70.01,
65.96, 65.85, 63.27, 38.32. ESI-MS: m/z 373 (2%, Mþ, C17
H
17
Cl2NO4),
37
37
35
371 (13%, Mþ, C17
H H Cl2NO4), 279
Cl35ClNO4), 369 (20%, Mþ, C20
17 21
(11), 277 (16), 241 (32), 214 (100), 180 (16). IR nmax (cmꢁ1): 3331,
concentration range from 0.5 to 50.0 mg/mL of each analyte, and
2951, 1720, 1504, 1451, 907, 728.
used to prepare calibration plots for all analytes. Validation data
for the HPLC analytical method are given in the Supplementary
Information.
Hydrolysis Reaction Solutions
The aqueous reaction solutions used in kinetic studies were
hydrochloric acid (pH 1.0, 1.5, and 2.0), citrate buffer (0.05 M; pH
3.0-5.0), phosphate buffer (0.067 M; pH 6.0-8.0), and borate buffer
(pH 9.0). Each solution was made with Milli-Q water and adjusted
to isotonicity with sodium chloride. The pH values of all reaction
solutions were measured with a calibrated high-precision pH meter
(Orion 520Aþ Advanced pH/mV/RmV/ORP meter; Thermo Electron
Corporation, CA). Calibration used the two-point protocol with pH
4 and 7 standards.
Results
Determination of the kinetic order and pseudo-first-order rate
constants for loss of DCF-MG (3a,b) and DCF-PHB-MG
In all buffered solutions, at constant pH and temperature, the
loss of DCF-MG (3a,b) and DCF-PHB-MG (3c) was described by first-
order kinetics. The observed pseudo-first-order rate constants, kobs
for disappearance of DCF-MG (3a,b) were calculated from the
slopes of plots of ln(concentration remaining) of the monomer
versus time according to Eq. (1).
,
Kinetic Procedure
Because of poor water solubility, 3a,b and 3c were first dissolved
in ACN, and then aliquots of each stock solution were carefully
ln½Ctꢃ ¼ ln½C0ꢃ ꢁ kobst
(1)
where the initial concentration is [C0], and [Ct] was the monomer
concentration as the sum of the concentration of the two possible
regioisomers [Ct] ¼ [3a] þ [3b], remaining at time, t. All plots were
linear over the duration of each study and kobs values were obtained
by linear regression, giving the kinetic data reported (see Tables S8
and S9 in Supplementary Information).
diluted with the hydrolysis media [(3a,b), 150 :
mM, 10% (v/v); 3c,
110 : M, 20% (v/v)]. To examine the effect of ACN on reactivity,
m
solutions of (3a,b) and (3c) at apparent pH 7.4 with varying con-
centrations of ACN (1%-30%, v/v) in the reaction medium were
degraded at 37ꢀC. The stabilities of the two main degradation
products (1) and (4) were also examined separately (pH 7.4, 37ꢀC).
All reactions were conducted in the dark, to avoid the possibility of
photolytic dehydrochlorination reactions, which have been
observed with closely similar chloroaryl NSAID structures, for
example, meclofenamic acid.22 Each buffered solution (10 mL) was
contained in a 20-mL screw cap glass vial with a magnetic stirring
bar. The capped vials were placed on a multiposition magnetic
stirrer (VARIOMAG™, TELESYSTEM 60 with TELEMODUL 20
controller; Thermo Electron LED GmbH, Osteroden, Germany),
which was submerged in a constant-temperature water bath
regulated at 37.0 0.5ꢀC. The pH values were checked at the end of
each experiment; no significant change was observed.
pH Rate Profiles
The loss of DCF-MG (3a,b) and DCF-PHB-MG (3c) was studied
over the pH range 1.0-8.0. The observed pseudo-first-order rate
constants (kobs) for loss of DCF esters (see Tables S8 and S9,
Supplementary Information) were plotted as pH rate profiles
(Fig. 1).
The profiles in Figure 1 allowed extraction of the second-order
rate constants for specific acid and specific base catalysis, as well
as the pseudo-first-order rate constant for uncatalyzed sponta-
neous hydrolysis, as shown in Table 1. A V-shaped pH rate profile
of DCF-MG (3a,b) loss over the pH range 1.0-8.0 was observed,
with maximum stability at about pH 4.7, whereas a U-shaped
profile was observed for loss of DCF-PHB-MG (3c) over the same
pH range, with maximum stability within the pH range 3.0-4.0. A
Transesterification at Low pH
Two individual kinetic experiments were performed at pH
1.0 to quantify the rate of ester hydrolysis for DCF-1- and

776
F. Wang et al. / Journal of Pharmaceutical Sciences 105 (2016) 773e785
significant change to the UV spectrum of DCF-MG. No significant
difference was seen between spectra in the range pH 1.0-3.0 (data
not shown), which suggested that no change in ionization state of
the N-H group occurred, and hence no effect on reaction kinetics
could be attributed to that cause.
Hydrolysis and Transesterification
The HPLC separation procedures described above permitted
baseline separation of all analytes shown in Scheme 1, including
the two isomers of DCF-MG (see Supplementary Information). For
DCF-MG, transesterification (acyl migration) involving the two
isomers, DCF-1-monoglyceride (3a) and DCF-2-monoglyceride
(3b), influence the hydrolysis kinetics of the ester. Trans-
esterification of the corresponding aryl isomers would be highly
unlikely to affect the hydrolysis rate of DCF-PHB-MG (3c), as the
isomers are remote from the reaction center and do not involve
the activated ester bond. Unless otherwise stated, all experimental
solutions were prepared from a common batch of DCF-MG that
contained 10%-15% of the 1-monoglyceride (3a), and corre-
spondingly 85%-90% of the 2-monoglyceride (3b), based on 1H
NMR spectra. However, transesterification was observed in buff-
ered solution over a wide pH range; this process was complete
within a few hours at neutral or higher pH values. In mildly acid
pH values (pH 3-5), transesterification was much slower, but was
found to be complete within approximately 350 h. Trans-
esterification gave formation of an equilibrium mixture of the
monoglycerides, favoring the 1-monoglyceride in a ratio of
approximately 8-9:1. At low pH (<3.0), the transesterification rate
was slow with respect to the rate of monomer degradation, such
that an approximate rate of ester hydrolysis for the individual
isomers could be measured. At pH 1.0, the observed loss of the
DCF-1-MG (3a) followed pseudo-first-order kinetics with a rate
constant of approximately k ¼ 0.0308 hꢁ1. The equivalent rate
Figure 1. log kobs pH profiles for loss of DCF-MG (3a,b) (,) and DCF-PHB-MG (3c) (D)
in aqueous buffers at 37.0ꢀC. Note data for DCF-MG (3a,b) were determined on the loss
of combined DCF-MG regioisomers [Ct] ¼ [3a] þ [3b], not for an individual regioisomer
and thus do not represent individual rate constants for hydrolysis of individual isomers
(see text).
small discontinuity in the acid catalyzed region of the DCF-MG
(3a,b) pH rate profile was observed between pH 2.0 and 3.0.
Linear regression equations for the pH ranges 1.0-2.0 and 3.0-4.5
were log kobs ¼ ꢁ0.9985pH þ 0.48 and log kobs ¼ ꢁ1.0014pH þ
0.15, respectively. The Fisher F-ratios for the two equations sug-
gested that their slopes were not statistically different from each
other, or from the theoretical value of ꢁ1.00. Further regression
equations were then estimated for the same two pH ranges,
with the gradient of each equation constrained to ꢁ1.000,
consistent with the first-order rate equation for specific acid
catalysis (Eq. (2)).
constant for the loss of DCF-2-MG (3b) was k ¼ ꢁ0.0969 hꢁ1
.
However, these rate constants are hybrid rate constants and
although they represent the loss of the monomer, they incorpo-
rate the combined loss from differing hydrolysis rates of each
individual regioisomer, transesterification, and possible contri-
bution from lactam formation. At pH 7.4, the equivalent hybrid
rate constants for loss of the two individual DCF-MGs were 0.0123
and 0.0102 hꢁ1, respectively (see Supplementary Information
Fig. S7).
log kobs ¼ log kH ꢁ pH
(2)
where kH is the specific second-order acid-catalyzed rate constant.
A Student's t-test (two-sided) of the difference (0.43) in the kH
values over the pH range 1.0-3.0, because of the discontinuity, was
found to be statistically significant at the 5% level.
A common explanation for a discontinuity such as that in the
acid-catalyzed region of the pH rate profile is a change in ionization
state of DCF-MG (3a,b). The N-H group of either DCF (1) or its
monomers may function as a very weak base, and thus form a
conjugate acid at very low pH. Any such change should result in a
At pH 1.0, contributions from acyl migration and ester hydrolysis
on the overall disappearance of the 1- and 2-monoglyceride are
evident.
A
pseudo-first-order reversible, parallel, and consecutive
process may describe the change in monomer concentration
(Scheme 2).
Table 1
Summary of Specific Catalytic and Spontaneous Rate Constants for Loss of Diclofenac Esters
Diclofenac Ester
kH (Mꢁ1 hꢁ1
)
kOH (Mꢁ1 hꢁ1
2.18 ꢂ 104
1.40 ꢂ 106
)
kH2O (hꢁ1
)
pHmax
4.7
<pH 2.0 (A)
>pH 2.0 (B)
DCF-MGa
3a,b
DCF-PHB-MG
3c
0.331b
0.459c
0.708b
0.450c
~0
0.261
1.32 ꢂ 10ꢁ3
3.0-4.0
For DCF-MG (3a,b), (A) and (B) refer to the specific acid-catalyzed rate constants for the region below the discontinuity (A; <pH ¼ 2.0) and above the discontinuity (B; pH > 2.0).
The pH value or pH range for maximum stability of (II) and (III) is also shown in the Table.
a
Rate constants reported in Table 1 (and Supplementary Information Table S8) for DCF-MG (3a,b) were determined based on the loss of the combined sum of the two
regioisomers DCF-1-MG (3a) and DCF-2-MG (3b), not for an individual regioisomer.
b
Slopes of the regression equations fixed at ꢁ1.00.
c
Slopes of the regression equations allowed to optimize.

F. Wang et al. / Journal of Pharmaceutical Sciences 105 (2016) 773e785
777
Scheme 1. Preparation of species used in this study. Reagents and conditions: (a) R1OH, DCC, DMAP, 0ꢀC, 2 h. (b) a: AcOH, 80ꢀC, 5 h; or b and c: Pd/C, H2, 1 atm, 16 h.
The rate equations derived from Scheme 2 are shown below:
relevant experiment number. Varying the molar ratio of 3a to 3b in
the two experiments allowed the rates of transesterification be-
tween isomers to be investigated.
The values of kf, kr, k1, and k2 were estimated at pH 1.0 by fitting
the differential equations to experimental data using a software
package (Table 2).
The estimated rate constants at pH 1.0 were used to simulate the
concentration versus time profiles of 3a and 3b in two experiments.
Experiment 1 had a starting mole ratio of 3a to 3b of 10:90, whereas
Experiment 2 was the opposite with a starting mole ratio of 3a to
3b of 90:10 (Fig. 2). Excellent correlation was observed between the
simulated curves and the experimental values (r2 values > 0.99)
(see Tables S10-S12 in Supplementary Information for additional
statistical analysis). At pH 1.0, the estimated rate constant for
disappearance of the 1-position isomer (3a) was only about 37%
that of the 2-position (3b) isomer.
d½3a1ꢃ
¼ kf ꢂ ½3b1ꢃ ꢁ ðkr þ k1Þ ꢂ ½3a1ꢃ
(3)
(4)
(5)
(6)
dt
ꢀ
ꢁ
d½3b1ꢃ
dt
¼ kr ꢂ ½3a1ꢃ ꢁ kf þ k2 ꢂ ½3b1ꢃ
d½3a2ꢃ
dt
¼ kf ꢂ ½3b2ꢃ ꢁ ðkr þ k1Þ ꢂ ½3a2ꢃ
ꢀ
ꢁ
d½3b2ꢃ
dt
¼ kr ꢂ ½3a2ꢃ ꢁ kf þ k2 ꢂ ½3b2ꢃ
where [3ax] and [3bx] are the concentrations of DCF-1- and
2-monoglycerides, respectively, and the subscript identifies the

778
F. Wang et al. / Journal of Pharmaceutical Sciences 105 (2016) 773e785
and lactam (4), were simultaneously determined as a function of
time, along with those for the monomers.
Figure 4 shows the pH-dependent mole fractions of the products,
DCF (1) and lactam (4), with respect to DCF-PHB-MG (3c) when
exposed to hydrolytic conditions (37ꢀC), and data for the time-
dependent simultaneously measured concentrations of DCF-PHB-
MG (3c), DCF (1), and lactam (4) at pHapp 4.0. Generally, formation
of lactam (4) was proportionally more extensive than formation of
DCF (1). At pH values below 3.0 and above 6.0, formation of lactam
(4) was more extensive than formation of DCF (1) from degradation
of DCF-PHB-MG (3c). By comparison, within the pH range 3.0-4.0,
DCF (1) accounted for more than 91% of the identified degradation
products, or 74% of the DCF-PHB-MG (3c). Mass balance between
loss of the monomer and formation of the two main products DCF (1)
and lactam (4) was not achieved for DCF-PHB-MG (3c) hydrolysis in
both acid and neutral conditions, yet approximately 90% of identified
analytes was recovered after complete loss of the substrate. A similar
relationship was observed with DCF-MG (3a,b), with proportionally
more lactam at pH values less than 4.0 and greater than 6.0. There
also appeared to be mass balance between loss of monomer and the
formation of the two products, DCF (1) and lactam (4), for DCF-MGs
(3a,b) in acid conditions. However, at pH 7.4, mass balance was not
achieved, with approximately 8-10 mole % loss of total substrate
during the experiment.
Scheme 2. Acyl migration between structural isomers, 3a and 3b, and degradation of
each species to diclofenac/diclofenac lactam at pH 1.0. Rate constants kf and kr describe
transesterification between isomers; k1 and k2 are the pseudo-first-order rate con-
stants describing the degradation of 3a and 3b, respectively.
Effect of ACN Cosolvent on Reactivity
Because of the low aqueous solubility of DCF (1),23 and the DCF
monomers (3a,b) and (3c), ACN was used as a cosolvent in the
reaction media to ensure complete dissolution at sufficient con-
centration. Thus, the effect of ACN content on reactivity was
examined to better understand the underlying mechanism and the
behavior of each monomer in the absence of ACN. To quantitatively
show the dependence of the hydrolysis reaction on the dielectric
constant (ε) of the medium, ln k was plotted as a linear function of
1/ε at the experimental temperature T (K),24 in accordance with the
following equation:
Although the data indicated that DCF (1) and lactam (4), in
varying proportions, were the main products from hydrolysis of
both DCF-2-MGs (3a,b) and DCF-PHB-MG (3c), there appeared to
be unidentified minor products.
The stabilities of DCF (1) and lactam (4) were examined indi-
vidually to consider the possibility of interconversion, or alternate
side reactions. There are a number of reports on the stability of DCF
(1) in acidic media.26,27 Hence, the reactivity of DCF (1) and lactam
(4) in 0.01 and 0.1 M NaOH was examined, as well as in 0.01 and 0.1
M HCl aqueous solutions. The results (Fig. 5) are summarized in
Table 3. DCF (1) was found to be unstable, particularly in acidic
solutions, as shown by the time-dependent concentration plot in
Figure 5a, suggesting direct formation of lactam (4) from DCF (1).
Conversely, lactam (4) appeared to be unstable, in basic solutions,
as shown by the time-dependent concentration plot in Figure 5b,
with the possibility for DCF (1) to also be formed from (4).
The data in Table 3 clearly shows the high susceptibility of DCF
(1) to cyclization and dehydration in acid solutions, forming the
lactam (4) at a rate proportional to [Hþ], consistent with the find-
ings of Larsen and Bundgaard,26 whereas the corresponding reac-
tion under alkaline conditions was far slower, with the rate not
proportional to [OHꢁ]. The data also show that the lactam was
unstable in neutral and basic conditions but very stable with no
sign of any degradation in acidic conditions (0.01 and 0.1 M HCl).
ZAm cos q
skTr2
ln k ¼ ln kε¼∞
þ
(7)
where kε ¼ ∞ is the rate constant at infinite dielectric constant, ZA is
the ionic charge of the attacking nucleophilic species (OHꢁanion),
m
is the molecular dipole moment, r is the shortest ion dipole
distance, and
k is Boltzmann's constant. ε was calculated ACN
and water,25 without taking into account the concentration of
phosphate buffer species. Figure 3 shows the dependency of the
observed first-order rate constants, kobs, on ACN concentration and
corresponding dielectric constant, ε. The negative slopes of the
plots were consistent with Eq. (7).
Mass Balance and Product Analysis of DCF-MGs (3a,b) and DCF-
PHB-MG (3c) Hydrolysis
Discussion
The two main end products from hydrolysis of DCF-MGs (3a,b)
and DCF-PHB-MG (3c), as identified and quantified by HPLC refer-
enced to the authentic compounds, were found to be DCF (1) and its
lactam (4); p-hydroxybenzoic acid was observed in the solvent
front. The concentrations of the two hydrolysis products, DCF (1)
Kinetics of DCF-MGs (3a,b) and DCF-PHB-MG (3c)
The rate of degradation of both DCF-MGs (3a,b) and DCF-PHB-
MG (3c) was found to follow pseudo-first-order kinetics at the
concentrations studied in the pH range 1.0-8.0. Both species were
subject to specific acid and base catalysis in this pH range. The
observed first-order rate constants were consistent with Eq. (8).
Table 2
Estimated Pseudo First-Order Rate Constants Describing the Transesterification/
Hydrolysis of 3a and 3b at pH 1.0
h
i
h
i
kobs ¼ kH Hþ þ ks þ kOH OHꢁ
(8)
Parameter
Estimated Rate Constant (10ꢁ5 hꢁ1
)
Standard Error (10ꢁ5 hꢁ1
)
kf
4450
506
2820
7630
90.1
13.0
21.7
where kH[Hþ] is the contribution from acid catalysis, ks is the rate
constant for spontaneous water attack, and kOH[OHꢁ] is the
contribution from base catalysis.
kr
k1
k2
125.2

F. Wang et al. / Journal of Pharmaceutical Sciences 105 (2016) 773e785
779
Figure 2. Concentration versus time profile describing the acyl migration and hydrolysis of DCF-MG isomers at 37.0ꢀC and pH 1.0. (a) Experiment 1: starting mole ratio 3a (-) to 3b
(,) of 10:90. (b) Experiment 2: starting mole ratio of 3a (C) to 3b (B) of 90:10. Squares and circles represent experimentally generated data, whereas lines represent data
simulated from nonlinear regression analysis.

780
F. Wang et al. / Journal of Pharmaceutical Sciences 105 (2016) 773e785
Figure 4. (a) Plot of the molar fraction of the products derived from DCF-PHB-MG (3c)
as a function of pH. The mole fraction is with respect to the starting monomer DCF-
PHB-MG (3c). (b) Mass balance plot for the hydrolysis of DCF-PHB-MG (3c) as a
Figure 3. In pseudo-first-order rate constants for loss of DCF-MG (3a,b) (,) and DCF-
PHB-MG (3c) (D) in IPB (apparent pH 7.4) as a function of (a) 1/g or (b) % (v/v) ACN. The
trend line was extrapolated to 0% (v/v) ACN giving ln k ¼ ꢁ4.039 (k ¼ 1.76 ꢂ 10ꢁ2 hꢁ1
)
◊
function of time at pH 4.0. Legend: DCF (1) ; DCF-PHB-MG (3c)
D, lactam (4), B; sum
for DCF-MG (3a,b) and ln k ¼ ꢁ0.322 (k ¼ 7.25 ꢂ 10ꢁ1 hꢁ1) for DCF-PHB-MG (3c).
^
of (1), (3c), and (4), I.
In the case of DCF-MG (3a,b), the data were consistent with an
uncatalyzed (spontaneous) first-order rate constant, ks, of approx-
imately 0 and second-order catalytic rate constants, kH and kOH, of
OHꢁ catalyzed hydrolysis than to Hþ catalyzed hydrolysis.28 The
profile shapes were also consistent with the hydrolysis behavior
expected for the activated ester DCF-PHB-MG (3c) (U-shaped pro-
file), which has a good leaving group, the p-hydroxybenzoate
moiety, compared with the unactivated ester DCF-MGs (3a,b)
(V-shaped profile), where the anion of glycerol was a relatively
poor leaving group. It was notable that the specific acid-catalyzed
region for each of the two profiles were virtually superimposable,
whereas the specific base-catalyzed regions were defined by
pseudo-second-order specific rate constants that differed by about
two orders of magnitude, indicating that the acid-catalyzed reac-
tion mechanism was barely influenced by the presence of the 40-O-
aromatic leaving group, unlike the base-catalyzed reaction, which
was significantly more influenced by the activated ester.
0.450 and 2.18 ꢂ 104 Mꢁ1
h
ꢁ1, respectively. For DCF-PHB-MG (3c),
the data were consistent with an uncatalyzed first-order rate con-
stant, ks, of 1.32 ꢂ 10ꢁ3
h
ꢁ1, and second-order catalytic rate con-
stants, kH and kOH, of 0.261 and 1.40 ꢂ 106
M
ꢁ1 hꢁ1, respectively.
pH Rate Profiles
The pH rate profile terminal slopes for acid- and base-catalyzed
degradation were close to the theoretical ꢁ1 and þ1 values,
respectively, especially for the specific base-catalyzed reactions
(3a,b, 0.992; 3c, 1.011). The plots depicted in the pH rate pro-file
(Fig. 1) were consistent with hydrolysis of the ester group, as the
pH for maximum stability in each case was on the weakly acid side
of neutrality, showing greater susceptibility of the ester group to
Study of the acid-catalyzed side of the profile for DCF-MG (3a,b)
suggested that it consisted of two parallel sections, that is, pH < 2
(A) and pH 2-4.5 (B), with a small discontinuity in the profile.

F. Wang et al. / Journal of Pharmaceutical Sciences 105 (2016) 773e785
781
Extrapolation of section (A) intersected with the Y-axis at log
k ¼ ꢁ0.48, whereas extrapolation of section (B) intersected with the
Y-axis at log k ¼ ꢁ0.15. These extrapolated values corresponded to
the pseudo-second-order acid-catalyzed rate constants for hydro-
lysis of the protonated and deprotonated substrates with k0Hþ
¼
0.331 Mꢁ1 hꢁ1 and kHþ ¼ 0.708 Mꢁ1 hꢁ1. Further statistical analysis
suggested the discontinuity to be beyond experimental error.
There are a number of possible explanations for the disconti-
nuity in the acid-catalyzed region of the DCF-MG (3a,b) pH rate
profile: (1) a change in ionization state of DCF-MG with pKa cor-
responding to the median pH of the observed displacement, (2) a
medium effect, and (3) citrate buffer catalysis.
Diclofenac has two ionization sites, the carboxylic acid (pKa ¼
3.99) and the very weakly basic diaryl amine, for which the pKa value
has not been reported, but is expected to be less than 2, if not less
than 0.26 Spectrophotometric results from our laboratory (data not
shown) suggested that DCF-MG: (1) had no change in ionization
down to pH 1; and (2) had an acidic pKa value of greater than 9.5.
A possible medium effect is a change in the apparent pH during
the course of the hydrolysis. An unbuffered medium (HCl) was used
in the kinetic studies at pH 1.0, 1.5, and 2.0, which corresponded to
section (A) of the displacement. The maximum change in pH that
could arise from complete hydrolysis of (3a,b) at a starting con-
centration of 150
mM was only ꢁ0.0065 (i.e., pH 2.00 to 1.9935). The
observed discontinuity was too big to be explained by any change in
pH during the experiment, despite the unbuffered medium.
The effect of buffer species on the hydrolysis of esters and on the
stability of drug species in general is well recognized. The catalytic
effect of citric acid buffer in particular on ester hydrolysis over an
acidic pH range is also well documented. For instance, citric acid
buffer has shown a catalytic effect on the hydrolysis of a neuro-
muscular blocking candidate that contained two ester linkages.29 It
is highly possible that a catalytic contribution from the citric acid
buffer is responsible for the observed discontinuity in the pH rate
profile as it occurred in the pH region covered by the citrate buffer
(pH 3.0-5.5) and is close to pKa1 for citric acid (3.1). As the
discontinuity occurred in a region that was not of interest for
wound care application, we did not further investigate buffer
catalysis by examining changes in buffer concentration.
Effect of ACN on Monomer Reactivity
◊
Figure 5. Reactivity of (a) DCF (1) ( ) in 0.1 M HCl (10% ACN) and (b) lactam (4) (B) in
0.1 M NaOH (10% ACN). ln %molar DCF (1) along with %molar lactam (4) generated from
DCF degradation was plotted as a function of time in (a), whereas ln %molar lactam (4)
remaining and log peak area unidentified product (Ú) from degradation of lactam was
plotted as a function of time as shown in (b).
As shown in Figure 3, the pseudo-first-order rate constants for
monomer hydrolysis decreased with decreasing dielectric constant
of the medium.
Table 3
Rate Parameters for Reactivity of 1 and 4
Reactant
Medium
Rate Constant
Half-Life (h)
r2
Apparent Zero Order
First Order (hꢁ1
)
Diclofenac 1
0.1 M NaOH
0.01 M NaOH
0.1 M HCl
0.01 M HCl
0.1 M NaOH
0.01 M NaOH
IPB pH 9.0
IPB pH 8.0
IPB pH 7.4
0.1 M HCl
e
3.51 ꢂ 10ꢁ5
19,720
20,310
20.1
219.8
2.21
2.15
17.5
79.0
1468
∞
0.9427
0.7291
0.9997
0.9948
0.9917
0.9891
0.9902
0.9950
0.9727
N/A
e
3.41 ꢂ 10ꢁ5
e
3.44 ꢂ 10ꢁ2
e
3.15 ꢂ 10ꢁ3
Lactam 4
e
2.43 ꢂ 10ꢁ2a
4.64 ꢂ 10ꢁ6b
5.70 ꢂ 10ꢁ7
1.26 ꢂ 10ꢁ7
1.36 ꢂ 10ꢁ8
0
e
e
e
e
e
e
0.01 M HCl
0
∞
N/A
a
This rate constant is from the apparent second-order model, which fitted the data best.
b
An apparent zero-order model fitted the data best under these conditions, but statistical tests suggested that the data were probably fractional order between zero and first
orders.

782
F. Wang et al. / Journal of Pharmaceutical Sciences 105 (2016) 773e785
The negative effect of the ACN content on the hydrolysis rates at
1-monoglyceride (2820 ꢂ 10ꢁ5
h
ꢁ1). This result suggested that, at
pH 7.4 is consistent with two mechanistic effects: (1) reduction
in water activity with increasing ACN content directly altered
the pseudo-first-order reaction conditions; or (2) influence of
the changing dielectric constant on transition state formation. The
present data (Fig. 3) are consistent with the first of these expla-
nations. Further studies with additional organic cosolvents are
needed to more fully distinguish between the two possibilities.
low pH, other factors, such as stabilization of the positively charged
carbonyl oxygen by a lone pair containing neighboring group such
as the two proximal primary hydroxyl groups or intramolecular
general acid catalysis, may be involved.30-32
Product Distribution and Mass BalancedDCF-MGs (3a,b)
The overall reaction pathways for the two substrates, DCF-MG
(3a,b) and DCF-PHB-MG (3c), are shown in Scheme 3. Detailed
examination of the quantitative kinetic data showed that DCF (1)
and the lactam (4) appeared to be the sole hydrolysis products of
DCF-MG (3a,b) within the pH range 4.5-5.5, and were formed in
equimolar amounts. The observed mass balance between loss of
monomer and formation of these two products suggested that they
were the only products from the hydrolysis. Combined with
the observed absence of a lag phase for either product, such
results were consistent with a pair of simultaneous parallel re-
actions, A / B and A / C (not considering transesterification be-
tween regioisomers). As the pH was decreased, proportionally
more lactam was formed, until at low pH (<3.0) lactam was the
predominant product. Mass balance was also observed between
loss of reactant and product formation over the pH range 1.0-4.0,
and again no lag phase was observed for either product, but there
was some evidence for loss of DCF (1).
Transesterification
Rapid transesterification (acyl migration) was observed for both
monomers, leading to equilibrium between the 1- and 2-isomers,
favoring the 1-monoglyceride (3a) in a molar ratio of about 8-9:1.
This was reasonable, given that, first, from a statistical point of view,
there are two 1ꢀ hydroxyl groups that can form the 1-monoglyceride
(3a) from the 2-monoglyceride (3b), whereas the 1-monoglyceride
(3a) has only one 2ꢀ hydroxyl that can form the 2-monoglyceride.
Second, the 2-monoglyceride (3b) was considered to be more ste-
rically crowded than the 1-monoglyceride (3a) and thus will be less
likely to form.
In this study, we observed that transesterification between the
2-monoglyceride (3b) and the 1-monoglyceride (3a) occurred over
the whole pH range. The reaction was faster at higher pH and
favored formation of 1-monoglyceride. For instance, at pH 7.4,
equilibrium was reached within the first hour, with a (3a)-(3b) ratio
of 8.2:1 (from a ratio of 0.4:1 at time zero); however, at pH 1.0,
slower transesterification could be followed by HPLC analysis to
quantitatively determine the rate of change in concentration for
each monomer. At pH 1.0, the data for the loss of the DCF-2-MG (3b)
are nonlinear (Fig. 2; Supplementary Information Fig. S7a), sug-
gesting that there is some component of transesterification to the
1-monoglyceride followed by hydrolysis occurring along with ester
hydrolysis of the 2-monoglyceride. This effect is only observed at
low pH, where the rate of transesterification is slow enough with
respect to the rate of ester hydrolysis. At higher pH, such as pH 7.4,
the rate of transesterification is rapid and no interference from
transesterification is seen. Thus, the apparent rate constant for the
2-monoglyceride at low pH will reflect the contributions of both
transesterification and hydrolysis.
As shown in Table 3, the lactam was stable under acidic condi-
tions (0.1 and 0.01 M HCl), while DCF (1) formed lactam (4). Such
results were consistent with an A / B þ C reaction mechanism to
explain the formation of DCF (1) and lactam (4) from DCF-MG
(3a,b) with an additional B / C reaction to explain the dehydra-
tion of DCF (1) to lactam (4). In specific acid-catalyzed hydrolysis,
protonation of the carbonyl oxygen leads to increased polarization
of the carbonyl group, facilitating addition of a nucleophile, that is,
a H2O molecule. As the pH was increased (>6.0), the reaction
mechanism was complicated by the instability of lactam (4). Con-
trary to the observed stability/instability in acidic conditions, under
basic conditions, DCF (1) was found to be stable and the lactam (4)
unstable. It was clear from the data in Table 3 that the lactam did
not form in alkaline aqueous solutions through cyclization dehy-
dration of DCF (1). Instability of the lactam was observed at a pH as
low as 7.4 (t1/2 ¼ 51 days), and became more rapid at higher pH.
Surprisingly, DCF (1) was not even a minor product of the
degradation of 4 under these conditions. Furthermore, the degra-
dation of lactam (4) fitted an apparent zero-order process, except at
high pH; this suggested an unidentified lactam degradation prod-
uct at neutral to higher pH values. One possibility was that base-
catalyzed dimerization leading to oligomerization of the lactam
might occur, a possibility supported by the appearance of broad,
low-intensity signals at long retention times on the chromato-
At pH 1.0, first-order rate constants for transesterification and
regioisomer degradation were determined.
The rate of the reverse reaction (3a to 3b) was observed to be an
order of magnitude slower than the rate of the forward reaction
(kr ¼ 506 ꢂ 10ꢁ5 hꢁ1 and kf ¼ 4450 ꢂ 10ꢁ5
h
ꢁ1, respectively).
Estimated values for rate constants describing acyl migration
support observations that the equilibrium between the 1- and 2-
isomers favors the 1-monoglyceride.
An apparent equilibrium constant, K, defined by Eq. (9), was
calculated to be 8.8 at pH 1.0 from the rate constants described
above. Similarly, K was approximated at 8.2 at pH 8.0, where
transesterification was much faster than hydrolysis, and where
equilibrium was reached within 1 h.
grams.26 The
a-methylene group on the lactam was considered to
be very weakly acidic.
The apparent zero-order model for the loss of lactam (Table 3),
tending toward a second-order model at high [OHꢁ], suggested that
the degradation process required hydroxyl ions to promote
deprotonation of the lactam at the a-carbon. Consistent with the
½ð3aÞꢃeq
kf
K ¼
¼
(9)
observations reported here, the rate limitation at near neutral to
mildly alkaline pH is the rate that a proton can be removed from the
substrate, which is dependent on the [OHꢁ] and not on the
significantly higher [substrate], whereas at higher pH, the limita-
tion would be the [substrate], accounting for the conversion to a
second (or even higher) order kinetic model in 0.1 M NaOH. The
resulting condensation product dimer is expected to have an
½ð3bÞꢃeq kr
The rate of disappearance was observed to differ for 1- and 2-
monoglycerides. Typically, bulky functional groups near reaction
centers reduce reactivity by sterically hindering the approach of the
incoming nucleophile. As the 2-monoglyceride was considered to
be more sterically crowded than the 1-monoglyceride, its ester
bond was expected to be less susceptible to hydrolysis.
extremely weakly ionizable
a-methylene and a carbonyl; these
However, the estimated rate constant for ester hydrolysis of the
features will allow further condensations leading to oligomeric
products. This hypothesis is also consistent with the observed
2-monoglyceride (7630 ꢂ 10ꢁ5
h
ꢁ1) was faster than that of the

F. Wang et al. / Journal of Pharmaceutical Sciences 105 (2016) 773e785
783
Scheme 3. Reaction pathways of (a) DCF-MGs (3a and b) and (b) DCF-PHB-MG (3c), showing the relationship between formation of the two products, DCF (1) and lactam (4).

784
F. Wang et al. / Journal of Pharmaceutical Sciences 105 (2016) 773e785
broad, highly retained, low-intensity unidentified signals in the
chromatograms.
kinetics, with two products resulting from hydrolysis, DCF (1), and
the lactam (4). The analytical methods developed for this study are
able to differentiate and quantify both DCF (1) and lactam (4) in a
common sample, something that was not previously reported. The
pH rate profile of the DCF-MGs (3a,b) was V-shaped with specific
acid catalysis below pH 4.5 and specific base catalysis above pH 4.5.
In contrast, the pH rate profile of DCF-PHB-MG (3c) was U-shaped
and characterized by specific acid catalysis below pH 4.5, specific
base catalysis above pH 4.5, and with spontaneous (uncatalyzed)
hydrolysis most dominant in the range pH 4.0-4.5.
The extensive formation of the lactam (4) from these monomers
might be expected to present an impediment to the use of the
pendent drugepolymer matrix approach to delivery of diclofenac.
However, data not yet published (in preparation) has indicated that
release of DCF (1) from such polymers proceeded with the forma-
tion of essentially no lactam at all. It was concluded that the results
in this study support the concept of drug delivery using a pendent
drug polymer matrix approach.
The pH rate profiles shown in Figure 1 indicated clearly that the
hydrolysis rate constants for both the unactivated alkyl ester and
the activated aryl ester were highly pH dependent, at least up to pH
8.0, with slopes of þ1, in the alkaline range, in both cases. It was
unlikely that the linear relationships in the pH rate profiles above
pH 8 will deviate from those straight lines, at least up to a pH value
of 9, which is of interest in the wound care field. Even in literature
examples where there was deviation from the strictly þ1 slope of
the observed pH rate profiles, it was very rare for the rate of hy-
drolysis to decrease at higher pH values, and never with esters.
Hence, a pendent DPC with diclofenac as the drug payload is
considered to release drug rapidly in the more basic conditions of
an exudating wound, than in the more nearly neutral conditions of
a healing wound.
The results of this work indicated that sufficient drug could be
released into an exudating wound bed at therapeutically relevant
levels from a DCF-MG (3a,b) or a DCF-PHB-MG (3c) conjugated
polymer dressing. Most importantly, because DCF (1) is a NSAID,
the inflammatory phase of chronic wound etiology could be posi-
tively influenced. Additionally, such delivery would address both
the pain and itching associated with chronic wounds.17 The release
rate profile of both DCF ester compounds in the pH range of chronic
wound exudate (~8.0-8.9)17,19,34 is particularly well suited to the
task and it is consistent with the desired release profile from a final
DCF (1) dressing that is exposed to a wound environment. Current
studies are measuring the release rate of DCF-MG (3a,b) and DCF-
PHB-MG (3c) conjugated polymer material in buffer and simu-
lated wound exudate.
The quantitative kinetic data also showed that DCF (1) and the
lactam (4) were not the sole products from the degradation of DCF-
MG (3a,b) at pH greater than 6.5. The most likely explanation for
lack of mass balance above pH 6.5 was loss of the lactam (4). The
impact of lactam instability was most apparent in the mass balance
plots of pH 6.5 and 7.0, which suggested that under neutral con-
ditions, the effect of [OHꢁ] on the instability of lactam was much
greater than the effect on the hydrolysis of DCF-MG (3a,b). The
data were consistent with a parallel pair of reactions A / B and
A / C, to explain the formation of 1 and 4 from (3a,b), with an
additional B / D reaction to explain the loss of lactam (4) to an as
yet unidentified species. In specific base catalysis, the tetrahedral
intermediate was formed by the addition of a nucleophilic OHꢁ
ion.26 This reaction is irreversible in principle for esters, as the
carboxylate ion formed is deprotonated in basic solution and,
hence, is not susceptible to attack by the nucleophilic alcohol or
phenol.
Product Distribution and Mass BalancedDCF-PHB-MG (3c)
The quantitative data for DCF-PHB-MG (3c) showed more rapid
hydrolysis across all studied pH ranges. This result was expected, as
it was generally understood that the rate of ester hydrolysis was
influenced most by the stability of the leaving alcohol, as the
phenoxide; the greater the stability of the leaving alkoxide (as
measured by the pKa of its free acid), the faster the hydrolysis rate.
Consequently, rate constants for hydrolysis of esters follow the
order: aryl esters > alkyl esters. For example, phenyl acetate hy-
drolyses about 12 times faster than ethyl acetate.28 The reaction
mechanisms for the acid-catalyzed hydrolysis and base-catalyzed
hydrolysis of DCF-PHB-MG (3c) appeared to be the same as those
for DCF-MG (3a,b). One notable exception was at pH 4.0, where DCF
(1) was the predominant product from the hydrolysis of DCF-PHB-
MG (3c), with minimal lactam (4) formation. The data were
consistent with DCF (1) being the sole product from the sponta-
neous uncatalyzed hydrolysis of DCF-PHB-MG (3c).
Other studies by us (as yet unpublished) have shown that the
hydrolytic release of DCF (1) from pendent DCFePHBepolyur-
ethane polymer conjugates at pH 7.4 was accompanied by signifi-
cantly reduced formation of (4), compared with that observed for
the monomers. It would therefore seem likely that release of DCF
(1) from such conjugates was mediated by spontaneous hydrolysis
of the ester linkage between DCF (1) and the pendent p-hydrox-
ybenzoic acid linker.
Lactam formation arises when the lone pair of electrons on ni-
trogen in the free base form attacks the carbonyl carbon of diclo-
fenac or diclofenac esters. As the basic pKa value for DCF, a diaryl
amine, was expected to be very low (e.g., aniline pKa ¼ 4.87;
o-chloroaniline pKa ¼ 2.64; diphenylamine pKa ¼ 0.79), lactam
formation was expected across almost the entire pH range. Varia-
tions in rate or extent of lactam formation as a function of pH will
then arise as a result of pH effects at more remote loci from the
central nitrogen atom. In this context, the state of ionization of the
free acid product DCF was also important. This has a measured pKa
value of 3.9933 for the carboxylic acid group. The results of our
studies are consistent with this theory. Of note is that the pH cor-
responding to minimum lactam formation appeared to coincide
with the pKa value for the carboxylic acid group.
This work extends the fundamental knowledge of DCF (1)
reactivity and chemistry and provides the framework for producing
biocompatible wound dressings that are covalently modified with
diclofenac. The pH-dependent release profile of both DCF (1) esters
is fully compatible with the chronic wound microenvironment. In
addition, the synthetic success of forming both the monoglyceride
and p-hydroxybenzoate-monoglyceride moieties indicated that
this approach has broad applicability for the covalent attachment of
other small molecules and their subsequent incorporation into
wound dressings. One can then envisage a single dressing that
releases several small molecule drugs that interact with different
aspects of chronic wound physiology in a temporal fashion to effect
healing.
Acknowledgments
Conclusions
This research was supported under the Australian Research
Council’s Linkage Projects funding scheme (project number
LP100100894). This study was also supported financially by
The present study showed that aqueous degradation of both
DCF-MGs (3a,b) and DCF-PHB-MG (3c) followed pseudo-first-order

F. Wang et al. / Journal of Pharmaceutical Sciences 105 (2016) 773e785
785
15. Tammara VK, Narurkar MM, Crider AM, Khan MA. Morpholinoalkyl ester
prodrugs of diclofenac: synthesis, in vitro and in vivo evaluation. J Pharm Sci.
1994;83(5):644-648.
16. Cassano R, Trombino S, Ferrarelli T, et al. Synthesis, characterization, and anti-
inflammatory activity of diclofenac-bound cotton fibers. Biomacromolecules.
2010;11(7):1716-1720.
17. Tsukada K, Tokunaga K, Iwama T, Mishima Y. The pH changes of pressure ulcers
related to the healing process of wounds. Wounds. 1992;4(1):16-20.
18. Lambers H, Piessens S, Bloem A, Pronk H, Finkel P. Natural skin surface pH is on
average below 5, which is beneficial for its resident flora. Int J Cosmet Sci.
2006;28(5):359-370.
PolyActiva Pty. Ltd. and Kimberly-Clark Corporation. Monash
Institute of Pharmaceutical Sciences received research funding
from PolyActiva Pty. Ltd. for this study. The following authors are
employees of PolyActiva Pty. Ltd.: Dr. Russell Tait, Dr. Andrew
Donohue, Dr. Asha D'Souza, and Dr. Sarah Ng. Dr. Stephen Quirk is
an employee of Kimberly-Clark Corporation.
We would like to thank Professor Bradley D. Anderson, Univer-
sity of Kentucky, for invaluable discussion and insights.
19. Paul J. A cross-sectional study of chronic wound-related pain and itching. Os-
tomy Wound Manage. 2013;59(7):28-34.
20. Cutting KF. Wound exudate: composition and functions. Br J Community Nurs.
2003;8(9-suppl):S4-S9.
References
21. Shukla VK, Shukla D, Tiwary SK, Agrawal S, Rastogi A. Evaluation of pH
measurement as a method of wound assessment. J Wound Care. 2007;16(7):
291-294.
1. D'Souza AM, Donohue AC, Tait RJ, Graichen FHM, Ng SMY, Sulistio A. 2014.
PolymereNSAID conjugate. Patent WO2014000033 A1.
2. Bandarage UK, Chen L, Fang X, et al. Nitrosothiol esters of diclofenac: synthesis
and pharmacological characterization as gastrointestinal-sparing prodrugs.
J Med Chem. 2000;43(21):4005-4016.
3. Vane JR. Inhibition of prostaglandin synthesis as a mechanism of action for
aspirin-like drugs. Nat New Biol. 1971;231:232-235.
4. Gan TJ. Diclofenac: an update on its mechanism of action and safety profile.
Curr Med Res Opin. 2010;26(7):1715-1731.
5. Dutta NK, Annadurai S, Mazumdar K, et al. Potential management of resistant
microbial infections with a novel non-antibiotic: The anti-inflammatory drug
diclofenac sodium. Int J Antimicrob Agents. 2007;30(3):242-249.
6. Mazumdar K, Dastidar SG, Park JH, Dutta NK. The anti-inflammatory non-
antibiotic helper compound diclofenac: an antibacterial drug target. Eur J Clin
Microbiol Infect Dis. 2009;28(8):881-891.
7. Harris KL, Bainbridge NJ, Jordan NR, Sharpe JR. The effect of topical analgesics
on ex vivo skin growth and human keratinocyte and fibroblast behavior.
Wound Repair Regen. 2009;17(3):340-346.
22. Philip J, Szulczewski DH. Photolytic decomposition of N-(2,6-dichloro-m-tolyl)
anthranilic acid (meclofenamic acid). J Pharm Sci. 1973;62(9):1479-1482.
ꢀ
23. Llinas A, Burley JC, Box KJ, Glen RC, Goodman JM. Diclofenac solubility: Inde-
pendent determination of the intrinsic solubility of three crystal forms. J Med
Chem. 2007;50(5):979-983.
24. Amis ES. Coulomb's law and the quantitative interpretation of reaction rates, II.
J Chem Educ. 1953;30:351-353.
25. Jouyban A, Soltanpour S, Chan H-K. A simple relationship between dielectric
constant of mixed solvents with solvent composition and temperature. Int J
Pharm. 2004;269(2):353-360.
26. Larsen C, Bundgaard H. Kinetics of the acid-catalyzed cyclization of diclo-
fenac to an indolinone in aqueous solution. Arch Pharm Chemi Sci Ed.
1980;8:100-108.
27. Reddersen K, Heberer T. Formation of an artifact of diclofenac during acidic
extraction of environmental water samples.
221-226.
J Chromatogr A. 2003;1011:
8. Krischak GD, Augut P, Claes L, Kinzl L, Beck A. The effects of non-steroidal anti-
inflammatory drug application on incisional wound healing in rats. J Wound
Care. 2007;16(2):76-78.
28. Connors KA, Amidon GL, Stella VJ. Chemical Stability of Pharmaceuticals: A
Handbook for Pharmacists. 2nd ed. New York: Wiley; 1986.
29. Zhu H, Meserve K, Floyd A. Preformulation studies for an ultrashort-acting
neuromuscular blocking agent GW280430A. I. Buffer and cosolvent effects on
the solution stability. Drug Dev Ind Pharm. 2002;28(2):135-142.
30. Bundgaard H, Buur A, Chang S-C, Lee VHL. Prodrugs of timolol for improved
ocular delivery: Synthesis, hydrolysis kinetics and lipophilicity of various
timolol esters. Int J Pharm. 1986;33(1-3):15-26.
31. Gu L, Dunn J, Dvorak C. Preformulation prodrug researchdChemical and
enzymatic hydrolysis kinetics of the glycerol, glycolic acid and morpholino
ethyl ester derivatives of a developmental analgesic agent (RS-82917). Drug
Dev Ind Pharm. 1989;15(2):209-221.
32. Pech B, Duval O, Richomme P, Benoit J-P. A timolol prodrug for improved
ocular delivery: Synthesis, conformational study and hydrolysis of palmitoyl
timolol malonate. Int J Pharm. 1996;128(1-2):179-188.
33. Prankerd R. Critical Compilation of pKa Values for Pharmaceutical Substances. 1st
ed. Amsterdam, The Netherlands: Associated Press; 2007.
9. Demidova-Rice TN, Hamblin MR, Herman IM. Acute and impaired wound
healing: pathophysiology and current methods for drug delivery, part 1:
Normal and chronic wounds: Biology, causes, and approaches to care. Adv Skin
Wound Care. 2012;25(7):304-314.
10. Bandgar BP, Sarangdhar RJ, Ahamed FA, Viswakarma S. Synthesis, character-
ization, and biological evaluation of novel diclofenac prodrugs. J Med Chem.
2011;54(5):1202-1210.
11. Tabrizi MHN, Davaran S, Entezami AA. Synthesis of diclofenac polymeric pro-
drugs and their hydrolysis reactivity. Iranian Polym J. 1996;5(4):243-249.
12. Bonina F, Puglia C, Santagati NA, Saija A, Tomaino A, Tita B. Oligoethylene ester
derivatives of ketoprofen, naproxen and diclofenac as oral prodrugs: a phar-
macological evaluation. Pharmazie. 2002;57(8):552-555.
13. Bonina FP, Puglia C, Barbuzzi T, et al. In vitro and in vivo evaluation of poly-
oxyethylene esters as dermal prodrugs of ketoprofen, naproxen and diclofe-
nac. Eur J Pharm Sci. 2001;14(2):123-134.
34. Vandenkerkhof EG, Hopman WM, Carley ME, Kuhnke JL, Harrison MB. Leg ulcer
nursing care in the community: A prospective cohort study of the symptom of
pain. BMC Nurs. 2013;12:3.
14. Jilani JA, Pillai GK, Salem MS, Najib NM. Evaluation of hydroxylethyl esters
of mefenamic acid and diclofenac as prodrugs. Drug Dev Ind Pharm.
1997;23(3):319.