Applicability of Phenolic Acids as Effective Enhancers of Cocrystal Solubility of Methylxanthines

Article abstract of DOI:10.1021/acs.cgd.7b00121

Applicability of phenolic acids as potential cocrystal formers for methylxanthine derivatives was analyzed both in terms of cocrystallization probabilities and solubility advantage. The cocrystal formation abilities were evaluated using mixing enthalpy estimated within the conductor like screening model for real solvents (COSMO-RS) framework. The solubility improvement of potential cocrystals was estimated by formulation of the model relating experimental values to predicted solubilities. This enabled for ranking of potential cocrystals formers according to their solubility enhancement potential. According to the calculation results, a highly linear relationship (R² = 0.989) was found between estimated theophylline and caffeine cocrystal solubility values. It has been found that many phenolic acids, especially ones with several hydroxyl groups attached to phenyl ring, are the most promising candidates for cocrystallization with caffeine or theophylline. Experimental verification of the proposed protocol for caffeine and theophylline resulted in eight new molecular complexes, which were synthesized via a mechanochemical approach. All new solids were characterized using powder X-ray diffractometry and Fourier transform infrared spectroscopy combined with a attenuated total reflection technique.

Full text of DOI:10.1021/acs.cgd.7b00121

Article
pubs.acs.org/crystal
Applicability of Phenolic Acids as Effective Enhancers of Cocrystal
Solubility of Methylxanthines
Maciej Przybyłek,† Dorota Ziołkowska,‡ Karina Mroczynska,§ and Piotr Cysewski*,†
́
́
†Chair and Department of Physical Chemistry, Pharmacy Faculty, Collegium Medicum of Bydgoszcz, Nicolaus Copernicus University
́ ́
in Torun, Kurpinskiego 5, 85-950 Bydgoszcz, Poland
§
‡Faculty of Chemical Technology and Engineering and Research Laboratory, Faculty of Chemical Technology and Engineering,
University of Technology and Life Sciences in Bydgoszcz, Seminaryjna 3, 85-326 Bydgoszcz, Poland
S
* Supporting Information
ABSTRACT: Applicability of phenolic acids as potential
cocrystal formers for methylxanthine derivatives was analyzed
both in terms of cocrystallization probabilities and solubility
advantage. The cocrystal formation abilities were evaluated using
mixing enthalpy estimated within the conductor like screening
model for real solvents (COSMO-RS) framework. The solubility
improvement of potential cocrystals was estimated by for-
mulation of the model relating experimental values to predicted
solubilities. This enabled for ranking of potential cocrystals
formers according to their solubility enhancement potential.
According to the calculation results, a highly linear relationship (R2 = 0.989) was found between estimated theophylline and
caffeine cocrystal solubility values. It has been found that many phenolic acids, especially ones with several hydroxyl groups
attached to phenyl ring, are the most promising candidates for cocrystallization with caffeine or theophylline. Experimental
verification of the proposed protocol for caffeine and theophylline resulted in eight new molecular complexes, which were
synthesized via a mechanochemical approach. All new solids were characterized using powder X-ray diffractometry and Fourier
transform infrared spectroscopy combined with a attenuated total reflection technique.
the specifics of the particular system. In the case of poorly
soluble drugs, there are some difficulties in applying solution
cocrystallization methods coming from low concentrations of
dissolved components. However, this can be overcome by using
popular mechanochemical cogrinding methods. As it was
demonstrated in many papers,27−36 this approach was applied
also in the case of methylxanthines including theophylline and
caffeine. In this paper, the liquid-assisted cogrinding method
was utilized for synthesis of new cocrystals of these two popular
drugs.
Many theoretical methods of rational cocrystal former
selection have been developed and tested against experimental
data sets.17,19,37−41 Recent approaches based on the analysis of
coformers similarity were expressed in terms of the
intermolecular interactions38,42,43 and mixing enthalpy of
supercooled liquids37 under ambient conditions. The latter
approach is based on the conductor like screening model for
real solvents (COSMO-RS), developed by Klamt et al.,44
whereby the likelihood of cocrystal formationcan be evaluated
by the heat of mixing enthalpies (Hmix) analysis.41 The more
negative Hmix, the highest probability of cocrystal formation.
The application of cocrystallization similarity concept offers
INTRODUCTION
■
Pharmaceutically interesting substances are very often poorly or
very poorly soluble in water. This is a source of serious
problems, especially from the perspective of the route of drugs
administration and consequently their bioavailability. That is
why many methods have been invented for improving the
physicochemical properties of new forms of drugs. Among
them one of the most promising routes is multicomponent
crystals synthesis.1 There are many spectacular examples of
improvements in pharmaceutically relevant properties achieved
through cocrystallization such as solubility,2−4 dissolution
rate,5−9 mechanical compressibility, tabletability,10,11 stabil-
ity,12−14 or hygroscopicity.15,16 A promising class of cocrystal
formers are nutraceuticals, such as flavonoids, vitamins, and
phenolic acids. According to our experimental and theoretical
studies, the latter compounds were found to have a high affinity
for amides such as urea, benzamide, salicylamide, and
ethenzamide.17−19 Because of antioxidant activities, phenolic
acids are often added to food, pharmaceuticals, and cosmetics
in order to enhance their stability.20−23
There are many efficient cocrystallization methods24−26
which can be classified into two main groups: slow
thermodynamic methods and fast kinetic approaches. All
popular techniques of cocrystals preparation, namely, slurry
cocrystallization, solvent evaporation, and grinding have some
limitations. Therefore, the choice of method is dependent on
Received: January 25, 2017
Revised: February 28, 2017
Published: March 3, 2017
© XXXX American Chemical Society
A
DOI: 10.1021/acs.cgd.7b00121
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
very fast theoretical screening by inferring cocrystallization of
one compound based on similarities between chemical species.
This in turns allows for filling gaps in the lists of known
cocrystals without the necessity of performing experiments of
all possible combinations of coformers pairs.
The aim of this paper is two-fold. First, further experimental
confirmation of the transferability of cocrystal formation
potential in the case of caffeine and theophylline was
performed. Since not all potential cocrystals are interesting
from the practical application, apart from high cocrystallization
probability, the second criterion was imposed on cocrystals
screening results. In this part of the study phenolic acids were
selected as potentially efficient cocrystal formers. Thus, the
conjuncture of two criterions defines the complete cocrystal
screening procedure presented below and applied in the case of
methylxanthine derivatives.
devoted to ranking of coformers with high probability of
cocrystallization using criterions of predicted solubility
advantage.
Transferability of Cocrystallization Landscapes be-
tween Theophylline and Caffeine. Since structures of
caffeine and theophylline are quite analogous, it is reasonable to
expect that they can exhibit similar abilities to cocrystal
formation. Indeed, as it can be inferred from Figure 1, there is
MATERIALS AND METHODS
■
Chemicals Used for Cocrystals Synthesis. All chemicals were
used without purification, as received from suppliers. Analytical grade
active pharmaceutical ingredients were used, namely, theophylline
(CAS: 58-55-9) and caffeine (CAS: 58-08-2). Also the following
cocrystal formers were utilized, namely, 4-amino-2-hydroxybenzoic
acid (4-aminosalicylic acid, CAS: 65-49-6), 4-nitrophenol (CAS: 100-
02-7), 2-hydroxy-1-naphthoic acid (CAS: 2283-08-1), 3-hydroxy-2-
naphthoic acid (β-hydroxynaphtoic acid, CAS: 92-70-6), 4-hydroxy-3-
methoxycinnamic acid (ferulic acid, CAS: 537-98-4), 1,2,4,5-
benzenetetracarboxylic acid (pyromellitic acid, CAS: 89-05-4), 2-
fluorobenzoic acid (CAS: 445-29-4), 3,4-dihydroxycinnamic acid (3,4-
dihydroxybenzeneacrylic acid and caffeic acid, CAS: 331-39-5) as
supplied by Sigma-Aldrich. Methanol (99.9%) was purchased from
Avantor Performance Materials Poland (Gliwice, Poland).
Figure 1. Correlation between mixing enthalpies of caffeine and
theophylline. The four series corresponds to different sets of coformers
mixed in the liquid state under supercooled conditions.
Experimental Cocrystals Screening. The procedure taking
advantage of mechanochemical synthesis utilized 0.1 g of methyl-
xanthine (theophylline or caffeine) mixed together with coformer in a
molar ratio of 1:1 and milled in a porcelain mortar for an hour. During
this procedure 20 μL portions of methanol were added to the mixture
at intervals of 2 min. The obtained samples were analyzed using
powder X-ray diffraction (PXRD) and Fourier transform infrared
spectroscopy combined with an attenuated total reflection (ATR)
technique (FTIR-ATR). Diffraction patterns were recorded using
PW3050/60 goniometer equipped with Empyrean XRD tube Cu LFF
DK303072. The range of 2θ angle was from 5° to 40° with 0.02° step.
The obtained raw diffractograms were processed including Kα2
stripping, background subtraction, smoothing, and normalization with
the use of the Reflex module of the Material Studio 8.0 package.45
FTIR-ATR spectra were measured on a Bruker Alpha-PFT-IR
spectrometer equipped with diamond attenuated total reflectance
accessory.
Calculation Details. The geometries in both gas and condensed
phases of all compounds were optimized using BP-RI/TZVP scheme,
which was followed by σ-profiles computation by means of BP-RI/
TZVPD approach taking advantage of TURBOMOLE software.46
Prior to this stage conformational analysis of all compounds was
performed using Materials Studio 8.0.45 Postquantum-mechanical
miscibility and aqueous solubility analyses were performed with
COSMOtherm software47 using BP_TZVPD_FINE_C30_1601.ctd
parametrization. The likelihood of cocrystal formation was estimated
based on the working paradigm of the correspondence between
miscibility in the solid state and in liquids quantified by the mixing
enthalpy.
quite a linear relationship between values of mixing enthalpies
of these two methylxanthines calculated for a common set of
coformers. In Figure 1 there are overlaid three sets of data. The
first one comprises coformers which were collected after
searching of the Cambridge Structural Database48 (CSD) for
binary solids comprising compounds listed in Table 1. As a
Table 1. Experimental Cocrystal Screening Results
a
Performed for Caffeine and Theophylline
coformer
caffeine
ref 53
theophylline
ferulic acid
+
−0.98
+
−0.69
ref 54
−0.84
4-amino-2-hydroxybenzoic acid
4-nitrophenol
−1.25
+
TOPPNP55
b
−0.59
LAKXUS25
−0.75
KIGKOB56
−0.74
−0.42
+
b
2-hydroxy-1-naphthoic acid
3-hydroxy-2-naphthoic acid
1,2,4,5-benzenetetracarboxylic acid
2-fluorobenzoic acid
−0.52
+
b
−0.52
+
b
NINXAL57
−2.44
−1.78
+
b
AFERAG58
−0.78
−0.59
+
caffeic acid
ref 53
RESULTS AND DISCUSSION
■
−1.71
−1.23
The two-step procedure for comprehensive cocrystals screening
started from estimation of similarities between cocrystals
landscapes of caffeine and theophylline and was followed by
experimental synthesis of new solids. The final step was
a
Plus symbol indicates cocrystals synthesized in this study. Values of
b
Hmix (in kcal/mol) are provided for each pair. Cocrystal reported in
CSD.
B
DOI: 10.1021/acs.cgd.7b00121
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
Figure 2. PXRD and FTIR spectra recorded for 1:1 caffeine/4-amino-2-hydroxybenzoic acid mixture (a, b) and 1:1 theophylline/ferulic acid mixture
(c, d) obtained via liquid-assisted grinding.
result a set of 162 cocrystals formed with 126 coformers was
found. Caffeine was used for the synthesis of 91 cocrystals, and
67 binary solids comprised theophylline. The other methyl-
xanthines were rarely used since seven cocrystals were found for
pentoxiphylline, six for theobromine, five for etophylline, and
one for doxophylline. The cocrystals of other methylxanthines
have been not deposited in the CSD. Using the set of 126
coformers, all possible combinations of binary mixtures were
considered for which Hmix values were computed. The resulting
distributions were collected in Figure 1 as black open circles
characterizing caffeine and theophylline. The second and the
most comprehensive set of coformers comes from the EAFUS/
GRAS lists comprising 835 species. Among them one can find
many naturally occurring nutraceuticals like vitamins, flavo-
noids, polyphenols, and phenolic acids, which are regarded as
potentially interesting and pharmaceutically relevant cocrystal
ingredients.49−52 This set was used for ensuring that the
extension of a coformers list beyond the one used in the
experiment also leads to similar linear relationships. Despite a
confirmative answer to this question, there are many pairs for
which the cocrystallization probability seems to be quite low as
indicated by high Hmix values. It was documented40 that in such
a case the region of low predictability of mixing enthalpy
corresponds to the values of Hmix higherthan −0.17 kcal/mol
when the BP-TZVPD-FINE level of computation is used.
Contrary to the first two sets comprising many hypothetical
pairs, the third collection was restricted only to experimentally
synthesized binary solids. In Figure 1 red circles document
cocrystals found in the literature. Additionally our experiments
enriched the collection of cocrystals formed by caffeine and
theophylline as documented by green diamonds.
New Cocrystals Synthesis. Although enormous reports on
the caffeine and theophylline cocrystals were published, there
are several gaps in their cocrystallization landscapes. Therefore,
experimental cocrystal screening studies using a popular
mechanochemical approach were performed for several
exemplary mixtures including phenolic acids. The screening
procedure was based on filling the gaps in the cocrystallization
landscapes of caffeine and theophylline comprising known
examples from CSD and the literature. As a result of cogrinding
experiments, eight new cocrystals were identified. These results
along with the mixing enthalpy values and structurally similar
cocrystals are summarized in Table 1. The complete
documentation of theophylline and caffeine cocrystal screening
results are provided in the Figure 2 and Supporting Information
(Figures S1−S6). In the case of all examined systems, the
C
DOI: 10.1021/acs.cgd.7b00121
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
cocrystal phase can be identified by the appearance of several
new peaks on each diffraction pattern recorded for the
coground mixture. For instance, the caffeine-4-amino-2-
hydroxybenzoic acid cocrystal can be detected by two new
diffraction peaks of high intensity located at 2θ angle values
equal to 10.7° and 11.1° and by several small signals at 14.5°,
19.7°, 21.1°, and 26.6° which cannot be assigned to the known
forms of the pure components (Figure 2a). In the case of
theophylline−ferulic acid system, cocrystal formation can be
evidenced by the appearance of a new signals on the diffraction
pattern, located at 8.2°, 9.9°, 11.1°, 13.7°, and 27° (Figure 2c).
Additional confirmation related to the new hydrogen bonding
motifs between coformers is provided by the FTIR-ATR
spectra analysis of coground mixtures and pure components. As
one can see from Figure 2b,d and Supplementary Figures S1−
S6, in each case several absorption bands appeared on the
vibrational spectra of mixtures which cannot be assigned to the
cocrystal formers. This is associated with absorption bands
shifts that indicate formation of a new molecular complex
stabilized by the OH···N interactions. As one can see, a blue-
shift of absorption bands in the OH stretching region, namely,
from 3491 to 3381 cm−1 and from 3385 to 3249 cm−1 can be
observed for the 4-amino-2-hydroxybenzoic−caffeine system
(Figure 2b). On the other hand, in the case of the exemplary
spectra recorded for theophylline−ferulic acid system, a
characteristic ν(OH) absorption band red-shift from 3432 to
3550 cm−1 probably indicates formation of a new molecular
complex (Figure 2d).
Extension of Cocrystallization Similarities on Other
Methylxanthines. On the basis of the cocrystallization
transferability concept,37 the miscibility of a particular
compound with a coformer can be inferred from the miscibility
of a structurally similar compound with the same coformer.
This is a contestant with the well-known rule “similia similibus
solventur”. Therefore, when the cocrystallization potential of
theophylline with a certain class of coformers is known, one can
evaluate the ability of cocrystal formation of other methyl-
xanthines. In Figure 3 the relationship between Hmix values
calculated for different cocrystals of methylxanthines found in
the CSD database with respect to Hmix values characterizing
theophylline are presented. As it can be seen, the highest
miscibility expressed by the most negative Hmix values can be
observed for pentoxiphylline, while the lowest miscibility occurs
for theodrenaline. Nevertheless, the differences between Hmix
values calculated for the most stable cocrystals of theodrenaline
and pentoxiphylline are quite small, namely, 1.87 kcal/mol.
Noteworthy, in all cases the highest miscibility can be observed
for octafluorohexanedioic acid, while the lowest can be
observed for urea. Although these systems were sparingly
studied experimentally, it seems very plausible that many new
cocrystals might be obtained with similar coformers as for
theophylline.
Figure 3. Relative values of mixing enthalpies of methylxanthines with
respect to theophylline. The abscissa characterizes Hmix of theophylline
with a variety of coformers, while the ordinate holds corresponding
values of other methylxanthines, namely, (2) caffeine, (3) pentoxiphyl-
line, (4) theobromine, (5) etophylline, (6) doxophylline, (7)
dyphylline, (8) lisophylline, (9) pentiphylline, (10) proxyphylline,
(11) theodrenaline, and (12) 8-cyclopentyltheophylline. In brackets,
the number of cocrystals found in the CSD database, correlation
coefficient, and the slope values are given. Series are composed from
coformers found in cocrystals deposited in the CSD.48
(0)
(S)
μ
− μ − max(0, ΔGfus)
CC
CC
log SCC
=
υtotRT ln(10)
(1)
where Scc denotes drug solubility in cocrystal, ΔGfus is the
(S)
enthalpy of fusion, μ(C0C) and μ stand for chemical potentials of
CC
the binary system in a vacuum (0) and solvent (s), respectively.
Since, for cocrystals usually there are no available experimental
data of ΔGfus the absolute solubility values cannot be computed
directly. Nevertheless, as suggested by Abramov et al.,59 API
cocrystals can be ranked in terms of their solubilities using the
relationship between experimental solubility advantage and
theoretical drug solubility in cocrystals calculated assuming
zeroth contributions from ΔGfus. The solubility advantage SA is
typically defined according to the following formula:
SCC
Sdrug
SA =
(2)
where Sdrug denotes solubility of pure API. As it can be inferred
from eq 2, for a certain drug (when Sdrug = const.) a linear
relationship between log SA and log SCC can be obtained. This
shows that there is a possibility for formulating models based
on known experimental SA data and theoretical SCC values. The
relationships between calculated solubility values and exper-
imental data reported in the literature were presented in Figure
4 including methylxanthinesstudied here. The observed trends,
although based on a limited number of cases, consistently
exhibit a linear increase in experimentally determined log SA
with an increase of estimated values of log SCC. This fortunate
correspondence enables at least qualitative estimating of
solubility advantage in cases of not studied cocrystals.
Particularly it is interesting to rank phenolic acids as potential
cocrystals formers with methylxanthines studied here. The
selection of these compounds was not accidental, since
Theoretical Solubility Advantage Analysis of Phenolic
Acids Methylxanthines Cocrystals. The selection of
cocrystal former is not only restricted to ones fulfilling the
miscibility criterion but should be also driven by the
pharmaceutically relevant properties of new solid formulations.
Since solubility enhancement is one of the main challenges of
API improvement, theoretical analysis of solubility advantage
was performed at the final stage of this study. According to the
COSMOtherm approach, multicomponent crystals solubility
can be calculated using the following equation:
D
DOI: 10.1021/acs.cgd.7b00121
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
Figure 4. Relationships between experimental values of solubility
advantage, log SAexp and computed drug solubilities in cocrystal
expressed as log SCC. The following systems were analyzed: A -
carbamazepine with saccharin, nicotinamide, succinic acid, malonic
acid, oxalic acid, salicylic acid, or glutaric acid in water,2,78 B -
theophylline with nicotinamide or salicylic acid in water,2,78 C -
caffeine with malonic acid, glutaric acid, maleic acid, salicylic acid or 1-
hydroxy-2-naphthoic acid in water,79,53 D, E, and F carbamazepine
with glutaric acid, nicotinamide, saccharin, and succinic acid in 2-
propanol, ethyl acetate, and ethanol, respectively.2,78
Figure 5. Correlation between estimated solubility values of caffeine
and theophylline in the case of cocrystallization with the same
coformers in a 1:1 stoichiometry.
which resembles the order of pure drug solubility in water. This
is also consistent with observations made by Good.78 In order
to rank the coformers in terms of their solubility advantage
potential, the theoretical values of log SCC were calculated for
hypothetical 1:1 cocrystals of theophylline and caffeine with
several classes of cocrystal formers. The presented results in
Figure 5 document a strong correlation between log SCC
distributions obtained for both methylxanthines.
phenolic acids and methylxanthines occur together in many
natural food products with well proven beneficial properties for
human health like yerba mate,60 cocoa,61 green coffee,62 and
tea.63 Compositions containing these compounds such as crude
caffeine were found to exhibit significant antioxidant, anti-
inflammatory, and glucose uptake enhancement activity.64
Apart from antioxidant activity, other important aspects of
phenolic acids and methylxanthines associated with solubility
and bioavailability enhancement can be distinguished. For
instance, natural herbal extracts rich in phenolic compounds
were found to improve pharmacokinetic properties of
theophylline.65 On the other hand, it has been proven that
addition of theophylline and caffeine contributes to the
bioavailability increase of many compounds including salicylic
acid.66 It is also worth mentioning that there are many
documented examples of solubility and dissolution behavior
enhancement achieved through crystallization of sparingly
soluble pharmaceuticals with phenolic acids.51,67−77 Previous
work37 and this study show that the miscibility of the unknown
system can be inferred from the similar known cases using
miscibility correlation plots (Figures 1 and 3), and, therefore, it
seems to be possible to predict solubility in a similar manner.
Indeed, there is a linear trend between caffeine and theophyl-
line cocrystals (Figure 5). The similarity of physicochemical
properties is the basis of all additive methods including well-
known Hansen’s and Hildebrandt’s solubility calculation
approach. It is understandable that structurally similar
compounds will exhibit similar solubility and changes in the
structure like removing and adding functional groups will have
an adequate and predictable influence on the solubility.
According to COSMO-RS calculations, the same rule applies
to multicomponent crystals.
In further analysis, it is assumed that the sensible solubility
advantage corresponds to SA > 2 that is associated with log SCC
> −2.8 in the case of caffeine cocrystals and log SCC > −1.9 in
case of the theophylline binary solids. These values were
estimated with the aid of trends documented in Figure 4 used
further for ranking of potential cocrystals formers. In Figure 5
the dotted line documents these threshold values. Despite the
fact that solubility advantage can differ by 1 order of magnitude,
it is expected that the same coformers can be used for
enhancing of solubility in water of both these methylxanthines.
It is interesting to notice that a sensible solubility advantage can
be observed for many phenolic acids, especially for ones rich in
carboxylic and phenolic groups. This can be explained by the
formation of stable complexes in water involving hydrogen
bonding between the COOH and OH group attached to the
benzene ring and imidazole nitrogen atoms. The highest values
of calculated solubility correspond to cocrystals formed by
cichoric acid, 3,4,5-trihydroxybenzoic acid (gallic acid), 3,4,5-
trihydroxycinnamic acid, and chlorogenic acid. All these four
coformers are expected to increase solubility of both caffeine
and theophylline cocrystals. The latter compound deserves
particular attention since it is present together with caffeine in
coffee. According to D’Amelio et al.80 caffeine forms very stable
complexes in the water with chlorogenic acid as evidenced by
the high association constant (K = 30 M−1) determined using
the nuclear magnetic resonance (NMR) titration method. It is
worth mentioning that formation of molecular complex in
aqueous solution was also observed using different spectro-
scopic methods in the case of caffeic acid and caffeine.81 These
facts explain the active role of phenolic acids in enhancing
cocrystal solubility.
As it can be inferred from Figure 4, the experimentally
proven solubility advantages of caffeine cocrystals are higher
compared to the analogous mixtures containing theophylline,
E
DOI: 10.1021/acs.cgd.7b00121
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
Apart from the mentioned phenolic acids, there are also
many more potential cocrystal formers worth considering. The
full list of performed screening is provided in Table 2.
3-methoxybenzoic acid are equal to −0.43 kcal/mol, while Hmix
values calculated for caffeine-3-hydroxybenzoic acid are equal to
−1.54 kcal/mol. These observations suggest that the presence
of hydroxyl groups attached to the benzene ring enhances the
ability of cocrystal formation due to the higher affinity of
coformer to methylxanthine and consequently altering of their
solubility. A similar effect can be observed in the case of
popular solubility enhancers, namely, cyclodextrins. Therefore,
it is understandable that cocrystals of the lowest solubilities are
formed by the phenolic acid precursor, cinnamic acid and O-
methylated phenolic acids, namely, 3,4-dimethoxybenzoic acid,
5-bromoferulic acid, and 3,4,5-trimethoxycinnamic acid.
Table 2. List of Coformers with Sensible Solubility
a
Advantage of Caffeine and Theophylline
log SCC
coformer
theophylline
caffeine
Phenolic Acids
cichoric acid
−0.36
−0.54
−0.85
−0.90
−1.03
−1.09
−1.15
−1.16
−1.01
−1.05
−1.31
−1.17
−1.48
−1.55
−1.55
−1.65
3,4,5-trihydroxybenzoic acid
3,4,5-trihydroxycinnamic acid
chlorogenic acid
CONCLUSIONS
■
caftaric acid
Theoretical cocrystal screening methods are useful for fast
selection of compounds which are the most probable to form a
molecular complex in the solid state. This approach is beneficial
from an economic viewpoint, since it can be used to minimize
the expensive experimental screening studies. Since theophyl-
line and caffeine have been extensively examined in terms of
cocrystal formation, it is possible to systematize this
information in order to formulate a model of cocrystallization
prediction based on the similarity concept. It is important,
however, to analyze compounds which are safe and which can
improve pharmaceutically relevant properties such solubility or
dissolution rate. In this paper, screening studies on the
methylxanthines’ miscibility with a variety of coformers were
performed including nutraceuticals belonging to GRAS and
EAFUS lists. The mixing enthalpy of liquids under supercooled
conditions estimated using COSMO-RS theory showed that
the majority of pharmaceutically acceptable coformers belong-
ing to EAFUS and GRAS lists exhibit relatively low affinity to
theophylline and caffeine. This narrows significantly the span of
potential coformers. Nevertheless, there are many other
compounds regarded as nutraceuticals which potentially can
be useful as isolated pure ingredients and compositions
obtained using modern extraction methods. In this study, a
promising class of cocrystal formers, namely, phenolic acids,
were analyzed in terms of their miscibility with methylxanthines
and aqueous solubility enhancement power. There are several
natural products like coffee, tea, and cocoa containing both
methylxanthines and phenolic acids. The wide interest of
research studies on these compounds is mainly associated with
their antioxidant activity. This study was aimed at a different
and also important aspect, which has not been explored in
detail elsewhere, namely, the solubility advantage evaluation of
a wide spectrum of methylxanthine−phenolic acid binary
mixtures. The absolute cocrystal solubility values calculation
poses practical difficulties related to the lack of enthalpy of
fusion ΔGfus values available in the literature. This cannot be
easily overcome. However, on the basis of the approach
reported by Abramov et al.,59 the coformers can be ranked in
terms of their solubility improvement potential using values
calculated assuming zeroth for ΔGfus. This allows for rational
selection of pharmaceutically interesting coformer. The
presented information relied on two-stage screening offering
suggestions of rational selection of coformers which not only
cocrystallize with the API but also will offer a sensible solubility
advantage. The application to other systems is only limited by
the knowledge of the solubility of some exemplary cocrystals
used in the validation step.
umbellic acid
3,4-dihydroxybenzoic acid
rosmarinic acid
Apliphatic Dicarboxylic Acids
oxalic acid
−0.79
−1.19
−0.95
−1.27
−1.46
−1.19
−1.41
−1.21
−1.56
−1.74
malonic acid
succinic acid
glutaric acid
adipic acid
EAFUS/GRAS Acids
phosphoric acid
sulfamic acid
0.00
−0.42
0.00
−0.54
−0.74
−0.26
−1.29
−1.65
1-hydroxyethylidene-1,1-diphosphonic acid
3-hydroxy-2-oxopropionic acid
hexanedioic acid
−1.05
−1.39
EAFUS/GRAS Alcohols
myricitrin
−0.63
−1.52
−1.52
−1.07
−1.78
−1.78
4-hydroxybenzenemethanol
4-hydroxybenzenemethanol
other
azodicarbonamide
−0.64
−0.84
a
Provided values characterize estimated water solubility of cocrystals.
It is worth mentioning that not all phenolic acids are to be
considered as effective solubility enhancers. Indeed, results of
our calculations suggest that cocrystals of methylxanthines with
salicylic acid exhibit relatively low solubility. This nicely
corresponds to experimental log SA values of caffeine−salicylic
acid and theophylline−salicylic acid cocrystals, which are as
small as 0.1753 and −0.40,2,78 respectively. The latter example
documents that cocrystallization of theophylline with salicylic
acid leads to the solubility reduction. This can be explained by
the relatively low affinity of methylxanthines to salicylic acid. It
is worth mentioning that salicylic acid−caffeine and salicylic
acid−theophylline crystal structures deposited in CSD under
refcodes XOBCAT82 and KIGLES83 are stabilized by a quite
weak synthon formed between COOH···N groups. In these
cocrystals, the OH group attached to the benzene ring is not
involved in the formation of strong intermolecular interactions
due to the competitive intramolecular interaction with the
COOH moiety. Another interesting observation can be made
by analyzing the number of hydroxyl groups attached to the
phenyl ring of coformers. Noteworthy phenolic acids
precursors with only one acidic functional group, namely,
benzoic acid and cinnamic acid, are less miscible with
theophylline and caffeine than their hydroxylated derivatives.
A similar observation can be made in the case of methylated
phenolic acids. For instance, Hmix values calculated for caffeine-
F
DOI: 10.1021/acs.cgd.7b00121
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
(21) Vangala, V. R.; Chow, P. S.; Tan, R. B. H. CrystEngComm 2011,
13 (3), 759−762.
(22) Jones, B. P. S.; Thigpen, D.; Morrison, J. L.; Richardson, A. P. J.
Am. Pharm. Assoc., Sci. Ed. 1956, 45 (4), 268−273.
(23) Ash, M.; Ash, I. Handbook of Preservatives; Synapse Information
Resources, Endicott: NY, 2004.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.cgd.7b00121.
■
S
PXRD patterns and FTIR spectra of 1:1 solid mixtures of
caffeine/4-nitrophenol, theophylline/2-hydroxy-1-naph-
toic, 1:1 theophylline/3-hydroxy-2-naphtoic acid mixture,
1:1 theophylline/1,2,4,5-benzenetetracarboxylic acid, 1:1
theophylline/2-fluorobenzoic acid, and 1:1 theophylline/
caffeic acid (PDF)
(24) Manin, A. N.; Voronin, A. P.; Drozd, K. V.; Manin, N. G.;
Bauer-Brandl, A.; Perlovich, G. L. Eur. J. Pharm. Sci. 2014, 65, 56−64.
̌
(25) Bucar, D.-K.; Henry, R. F.; Duerst, R. W.; Lou, X.; MacGillivray,
L. R.; Zhang, G. G. Z. J. Chem. Crystallogr. 2010, 40 (11), 933−939.
(26) Gagniere, E.; Mangin, D.; Puel, F.; Rivoire, A.; Monnier, O.;
̀
Garcia, E.; Klein, J. P. J. Cryst. Growth 2009, 311 (9), 2689−2695.
(27) Fischer, F.; Scholz, G.; Benemann, S.; Rademann, K.;
Emmerling, F. CrystEngComm 2014, 16 (35), 8272−8278.
(28) Shimono, K.; Kadota, K.; Tozuka, Y.; Shimosaka, A.; Shirakawa,
Y.; Hidaka, J. Eur. J. Pharm. Sci. 2015, 76, 217−224.
AUTHOR INFORMATION
Corresponding Author
*E-mail: piotr.cysewski@cm.umk.pl.
ORCID
Piotr Cysewski: 0000-0001-6220-2634
Notes
The authors declare no competing financial interest.
■
(29) Lu, J.; Rohani, S. J. Pharm. Sci. 2010, 99 (9), 4042−4047.
(30) Fischer, F.; Heidrich, A.; Greiser, S.; Benemann, S.; Rademann,
K.; Emmerling, F. Cryst. Growth Des. 2016, 16 (3), 1701−1707.
(31) Heiden, S.; Trobs, L.; Wenzel, K.-J.; Emmerling, F.
CrystEngComm 2012, 14 (16), 5128−5129.
̌ ̌ ́
(32) Karki, S.; Friscic, T.; Jones, W.; Motherwell, W. D. S. Mol.
Pharmaceutics 2007, 4 (3), 347−354.
ACKNOWLEDGMENTS
Technical assistance of Anna Ciesl
■
(33) Abourahma, H.; Urban, J. M.; Morozowich, N.; Chan, B.
CrystEngComm 2012, 14 (19), 6163.
́
in
ska (MD) and Tomasz
Jelin
́
(34) Fischer, F.; Scholz, G.; Batzdorf, L. CrystEngComm 2014, 17 (4),
2−4.
REFERENCES
(35) Fischer, F.; Lubjuhn, D.; Greiser, S.; Rademann, K.; Emmerling,
F. Cryst. Growth Des. 2016, 16 (10), 5843−5851.
(36) Hasa, D.; Miniussi, E.; Jones, W. Cryst. Growth Des. 2016, 16
(8), 4582−4588.
■
(1) Qiao, N.; Li, M.; Schlindwein, W.; Malek, N.; Davies, A.;
Trappitt, G. Int. J. Pharm. 2011, 419 (1), 1−11.
(2) Good, D. J.; Rodríguez-Hornedo, N. Cryst. Growth Des. 2009, 9
(5), 2252−2264.
(37) Cysewski, P. Struct. Chem. 2016, 27 (5), 1403−1412.
(38) Zhou, L.; Dodd, S.; Capacci-Daniel, C.; Garad, S.; Panicucci, R.;
Sethuraman, V. Eur. J. Pharm. Sci. 2016, 88, 191−201.
(39) Cysewski, P. New J. Chem. 2016, 40 (1), 187−194.
(40) Cysewski, P. J. Mol. Graphics Modell. 2016, 68, 23−28.
(41) Klamt, A. J. Cheminform. 2012, 4 (Suppl 1), O14.
(42) Mapp, L. K.; Coles, S. J.; Aitipamula, S. Cryst. Growth Des. 2017,
17, 163.
(3) Thakuria, R.; Delori, A.; Jones, W.; Lipert, M. P.; Roy, L.;
Rodríguez-Hornedo, N. Int. J. Pharm. 2013, 453 (1), 101−125.
(4) Lipert, M. P.; Rodríguez-Hornedo, N. Mol. Pharmaceutics 2015,
12 (10), 3535−3546.
(5) Huang, Y.; Zhang, B.; Gao, Y.; Zhang, J.; Shi, L. J. Pharm. Sci.
2014, 103 (8), 2330−2337.
(6) Chen, Y.; Li, L.; Yao, J.; Ma, Y. Y.; Chen, J. M.; Lu, T. B. Cryst.
Growth Des. 2016, 16 (5), 2923−2930.
(43) Corpinot, M. K.; Guo, R.; Tocher, D. A.; Buanz, A. B. M.;
(7) Gao, Y.; Zu, H.; Zhang, J. J. Pharm. Pharmacol. 2011, 63 (4),
483−490.
̌
Gaisford, S.; Price, S. L.; Bucar, D.-K. Cryst. Growth Des. 2017, 17 (2),
827−833.
(8) McNamara, D. P.; Childs, S. L.; Giordano, J.; Iarriccio, A.;
Cassidy, J.; Shet, M. S.; Mannion, R.; O’Donnell, E.; Park, A. Pharm.
Res. 2006, 23 (8), 1888−1897.
(44) Klamt, A. J. Phys. Chem. 1995, 99 (7), 2224−2235.
(45) Accelrys Materials Studio 7-Inc. 2014.
(46) Ahlrichs, R.; Bar, M.; Haser, M.; Horn, H.; Kolmel, C. Chem.
̈
̈
̈
(9) Remenar, J. F.; Peterson, M. L.; Stephens, P. W.; Zhang, Z.;
Zimenkov, Y.; Hickey, M. B. Mol. Pharmaceutics 2007, 4 (3), 386−400.
Phys. Lett. 1989, 162 (3), 165−169.
(47) Eckert, F.; Klamt, A. COSMOlogic Gm bH & Co. KG:
Leverkusen, 2014.
(48) Groom, C. R.; Bruno, I. J.; Lightfoot, M. P.; Ward, S. C. Acta
Crystallogr., Sect. B: Struct. Sci., Cryst. Eng. Mater. 2016, 72 (2), 171−
179.
̌ ̌ ́ ́
(10) Karki, S.; Friscic, T.; Fabian, L.; Laity, P. R.; Day, G. M.; Jones,
W. Adv. Mater. 2009, 21 (38−39), 3905−3909.
(11) Ahmed, H.; Shimpi, M. R.; Velaga, S. P. Drug Dev. Ind. Pharm.
2017, 43, 89−97.
(12) Reddy, L. S.; Bethune, S. J.; Kampf, J. W.; Rodríguez-Hornedo,
N. Cryst. Growth Des. 2009, 9 (1), 378−385.
(49) Sinha, A. S.; Maguire, A. R.; Lawrence, S. E. Cryst. Growth Des.
2015, 15 (2), 984−1009.
(13) Lin, R. Z.; Sun, P. J.; Tao, Q.; Yao, J.; Chen, J. M.; Lu, T. B. Eur.
J. Pharm. Sci. 2016, 85, 141−148.
(50) Sinha, A. S.; Rao Khandavilli, U. B.; O’Connor, E. L.; Deadman,
B. J.; Maguire, A. R.; Lawrence, S. CrystEngComm 2015, 17 (26),
4832−4841.
(14) Sarcevica, I.; Orola, L.; Veidis, M. V.; Podjava, A.; Belyakov, S.
Cryst. Growth Des. 2013, 13 (3), 1082−1090.
(51) Sekhon, B. RGUHS J. Pharm. Sci. 2012, 2 (2), 16−25.
(15) Chow, S. F.; Chen, M.; Shi, L.; Chow, A. H. L.; Sun, C. C.
Pharm. Res. 2012, 29 (7), 1854−1865.
̌ ́
(52) Veverka, M.; Dubaj, T.; Gallovic, J.; Jorík, V.; Veverkova, E.;
̌
́
Danihelova, M.; Simon, P. Monatsh. Chem. 2015, 146 (1), 99−109.
(16) Wang, Z. Z.; Chen, J. M.; Lu, T. B. Cryst. Growth Des. 2012, 12
(9), 4562−4566.
(53) Mukherjee, S.Crystal Engineering of Pharmaceutical Cocrystals.
MS Thesis, University of South Florida, 2011
(17) Przybyłek, M. I.; Zioł
Eur. J. Pharm. Sci. 2016, 85, 132−140.
(18) Przybyłek, M.; Ziołkowska, D.; Kobierski, M.; Mroczyn
Cysewski, P. J. Cryst. Growth 2016, 433, 128−138.
(19) Cysewski, P.; Przybyłek, M.; Ziołkowska, D.; Mroczyn
Mol. Model. 2016, 22 (5), 103.
́ ́
kowska, D.; Mroczynska, K.; Cysewski, P.
(54) Fernandes, J. A.; Sardo, M.; Mafra, L.; Choquesillo-Lazarte, D.;
Masciocchi, N. Cryst. Growth Des. 2015, 15 (8), 3674−3683.
(55) Aoki, K.; Ichikawa, T.; Koinuma, Y.; Iitaka, Y. Acta Crystallogr.,
Sect. B: Struct. Crystallogr. Cryst. Chem. 1978, 34 (7), 2333−2336.
(56) Bucar, D. K.; Henry, R. F.; Lou, X.; Duerst, R. W.; Borchardt, T.
̌
B.; MacGillivray, L. R.; Zhang, G. G. Z. Mol. Pharmaceutics 2007, 4
(3), 339−346.
́
́
́
́
(20) Maqsood, S.; Benjakul, S.; Shahidi, F. Crit. Rev. Food Sci. Nutr.
2013, 53 (2), 162−179.
G
DOI: 10.1021/acs.cgd.7b00121
Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design
Article
(57) Arman, H. D.; Tiekink, E. R. T. Acta Crystallogr., Sect. E: Struct.
Rep. Online 2013, 69 (9), o1443−o1443.
̌
(58) Bucar, D.-K.; Day, G. M.; Halasz, I.; Zhang, G. G. Z.; Sander, J.
R. G.; Reid, D. G.; MacGillivray, L. R.; Duer, M. J.; Jones, W. Chem.
Sci. 2013, 4 (12), 4417−4425.
(59) Abramov, Y. A.; Loschen, C.; Klamt, A. J. Pharm. Sci. 2012, 101
(10), 3687−3697.
́
(60) Vieira, M. A.; Maraschin, M.; Pagliosa, C. M.; Podesta, R.; de
Simas, K. N.; Rockenbach, I. I.; Amboni, R. D. de M. C.; Amante, E. R.
J. Food Sci. 2010, 75 (3), C280−C285.
(61) Leite, P. B.; Maciel, L. F.; Opretzka, L. C. F.; Soares, S. E.; da
Silva Bispo, E. Cienc. Agrotecnol. 2013, 37 (3), 244−250.
(62) Alonso-Salces, R. M.; Serra, F.; Reniero, F.; Heberger, K. J. Agric.
Food Chem. 2009, 57 (10), 4224−4235.
(63) Bi, W.; He, C.; Ma, Y.; Shen, J.; Zhang, L. H.; Peng, Y.; Xiao, P.
Acta Pharm. Sin. B 2016, 6 (2), 170−181.
(64) Chu, Y. F.; Chen, Y.; Brown, P. H.; Lyle, B. J.; Black, R. M.;
Cheng, I. H.; Ou, B.; Prior, R. L. Food Chem. 2012, 131 (2), 564−568.
(65) Al-Mohizea, A. M.; Raish, M.; Ahad, A.; Al-Jenoobi, F. I.; Alam,
M. A. J. Tradit. Chin. Med. 2015, 35 (5), 588−593.
(66) Beubler, E.; Lembeck, F. Naunyn-Schmiedeberg's Arch.
Pharmacol. 1976, 292 (1), 73−77.
(67) Aitipamula, S.; Wong, A. B. H.; Chow, P. S.; Tan, R. B. H.
CrystEngComm 2012, 14 (24), 8515−8524.
(68) Gadade, D. D.; Pekamwar, S. S.; Lahoti, S. R.; Patni, S. D.;
Sarode, M. C. Marmara Pharm. J. 2016, 21 (1), 78−88.
(69) Zhang, T.; Yang, Y.; Wang, H.; Sun, F.; Zhao, X.; Jia, J.; Liu, J.;
Guo, W.; Cui, X.; Gu, J.; Zhu, G. Cryst. Growth Des. 2013, 13 (12),
5261−5266.
(70) Zhang, T.; Yang, Y.; Zhao, X.; Jia, J.; Su, H.; He, H.; Gu, J.; Zhu,
G. CrystEngComm 2014, 16 (33), 7667.
(71) Vinesha, V.; Sevukarajan, M.; Rajalakshmi, R.; Chowdary, G. T.;
Haritha, K. Int. Res. J. Pharm. 2013, 4 (4), 218−223.
(72) Serrano, D. R.; Persoons, T.; D’Arcy, D. M.; Galiana, C.; Dea-
Ayuela, M. A.; Healy, A. M. Eur. J. Pharm. Sci. 2016, 89, 125−136.
(73) Arenas-García, J. I.; Herrera-Ruiz, D.; Morales-Rojas, H.; Hopfl,
̈
H. Eur. J. Pharm. Sci. 2017, 96, 299−308.
(74) Zhang, X.; Tian, Y.; Jia, J.; Zhang, T.; Zhu, G. J. Mol. Struct.
2016, 1108, 560−566.
(75) Nechipadappu, S. K.; Trivedi, D. R. Eur. J. Pharm. Sci. 2017, 96,
578−589.
(76) Hariprasad, V. M.; Nechipadappu, S. K.; Trivedi, D. R. Cryst.
Growth Des. 2016, 16 (8), 4473−4481.
(77) Yoshimura, M.; Miyake, M.; Kawato, T.; Bando, M.; Toda, M.;
Kato, Y.; Fukami, T.; Ozeki, T. Cryst. Growth Des. 2017, 17, 550.
(78) Good, D. J. Pharmaceutical Cocrystal Eutectic Analysis: Study of
Thermodynamic Stability, Solubility, and Phase Behavior. PhD Thesis,
University of Michigan, Ann Arbor, MI, 2010.
(79) Alsirawan, M. B.; Vangala, V.; Paradkar, A. Conference:
Fundamentals of solid forms; New insights and development, At
Burlington House, London, UK, 2016.
(80) D’Amelio, N.; Fontanive, L.; Uggeri, F.; Suggi-Liverani, F.;
Navarini, L. Food Biophys. 2009, 4 (4), 321−330.
(81) Belay, A.; Kim, H. K.; Hwang, Y.-H. Luminescence 2016, 31 (2),
565−572.
(82) Lu, E.; Rodríguez-Hornedo, N.; Suryanarayanan, R. CrystEng-
Comm 2008, 10 (6), 665.
(83) Childs, S. L.; Stahly, G. P.; Park, A. Mol. Pharmaceutics 2007, 4
(3), 323−338.
H
DOI: 10.1021/acs.cgd.7b00121
Cryst. Growth Des. XXXX, XXX, XXX−XXX