Anda di halaman 1dari 16

Didanosine Polymorphism in a Supercritical

Antisolvent Process

R. BETTINI,1 R. MENABENI,1 R. TOZZI,1 M.B. PRANZO,1 I. PASQUALI,1 M. R. CHIEROTTI,2


R. GOBETTO,2 L. PELLEGRINO2
1
Department of Pharmacy, University of Parma, Parma, Italy
2
Department of Chemistry I.F.M., University of Turin, Turin, Italy

Received 7 July 2009; revised 20 August 2009; accepted 26 August 2009


Published online 13 October 2009 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21962

ABSTRACT: Solid-state properties of active ingredients are crucial in pharmaceutical


development owing to their significant clinical and economical implications. In the
present work we investigated the solid-state properties and the solubility in water
of didanosine, DDI, re-crystallized from a dimethylsulfoxide solution using supercritical
CO2 as an antisolvent (SAS process) for comparison with the commercially available
drug product. We also applied modern solid-state NMR (SS NMR) techniques, namely
2D 1H DQ CRAMPS (Combined Rotation And Multiple Pulse Spectroscopy) and
1
H–13C on- and off-resonance CP (cross polarization) FSLG-HETCOR experiments,
known for providing reliable information about 1H–1H and 1H–13C intra- and inter-
molecular proximities, in order to address polymorphism issues arising from the crystal-
lization of a new form in the supercritical process. A new polymorph of didanosine was
obtained from the supercritical antisolvent process and characterized by means of 1D
and 2D multinuclear (1H, 13C, 15N) SS NMR. The particle size of the new crystal phase
was reduced by varying the antisolvent density through a pressure increase. The
structural differences between the commercial product and the SAS re-crystallized
DDI are highlighted by X-ray diffractometry and well described by solid-state NMR.
The carbon C6 13C chemical shift suggests that both commercial and re-crystallized
didanosine samples are in the enol form. The analysis of homo- and heteronuclear
proximities obtained by means of 2D NMR experiments shows that commercial and
SAS re-crystallized DDI possess very similar molecular conformation and hydrogen
bond network, but different packing. The new polymorph proved to be a metastable form
at ambient conditions, showing higher solubility in water and lower stability to
mechanical stress. ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm
Sci 99:1855–1870, 2010
Keywords: supercritical antisolvent crystallization; solid-state NMR; polymorphism;
didanosine

INTRODUCTION

Solid-state properties of active ingredients are


crucial in pharmaceutical development owing to
their significant clinical and economical implica-
Additional Supporting Information may be found in the tions.1–4 The discovery and isolation of different
online version of this article.
crystal forms of a new or already marketed drug
Correspondence to: R. Bettini (Telephone: 39-0521905089; Fax: 39-
0521905006; E-mail: bettini@unipr.it) represents a remarkable issue that can dramati-
Journal of Pharmaceutical Sciences, Vol. 99, 1855–1870 (2010) cally impact on both clinical efficacy (i.e., either by
ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association varying the drug dissolution rate and, therefore,

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010 1855


1856 BETTINI ET AL.

the bioavailability or by allowing the drug


administration through a new route) and indus-
trial production of the dosage forms. Sometimes
even small changes in the crystal structure afford
significant variation in drug dissolution rate or
stability.5
Modern technologies based on supercritical
fluids offer new perspectives for the design and
production of innovative materials6 as they afford
peculiar process conditions and offer the possi-
bility of avoiding or minimizing the use of organic
solvents.7
Supercritical fluids are nowadays proposed in
the pharmaceutical field as alternative ‘‘solvent’’
for a series of important operations such as
micron- and sub micron-sized particle production,
residual solvents removal and drug loading of
polymers.7 In all these cases, a great innovation Figure 1. Chemical structure of DDI.
potential exists for the production of new crystal-
line or amorphous phases with improved and
advantageous technological and biopharmaceuti-
cal characteristics and features.6 As opposed Didanosine (20 ,30 -dideoxyinosine), DDI (Fig. 1)
to traditional techniques for micron-sized is a reverse transcriptase inhibitor, effective
particle production (crystallization followed by against HIV and used in combination with
comminution), that may lead to cohesive, elec- other antiretroviral drugs as part of highly active
trically charged and poorly crystalline powders, antiretroviral therapy.
the supercritical fluid based processes have the Nevertheless, it presents some drawbacks such
potential for affording in one production step as tendency to hydrolyze at the acid pH of
noncohesive, free flowing and highly crystalline the stomach, low oral bioavailability, short
powders.8 biological half-life. These issues can be addressed
The production of microparticles by super- by site specific drug delivery systems14–17 or by
critical fluids can be achieved mainly by means exploiting possible alternatives to oral route.18–20
of three possible alternatives: (i) rapid expansion This approach may require the need for accurate
of a drug solution in a supercritical fluid,9 (ii) drug control of the physico-chemical properties of
re-crystallization by adding supercritical CO2 as the drug particles such as size, shape, crystalline
antisolvent,10–13 or (iii) drug precipitation in an purity and dissolution rate. No reports are
amorphous state or as nano-sized crystals on presently available on the solid-state chemistry
hydrophilic polymers.7 The crystallization of a (X-ray structure, polymorphism studies, etc.) of
drug from solutions in supercritical CO2 can this active principle.
exploit peculiar process conditions (e.g., pressure The aim of this work was to investigate the
and/or temperature variation, rate of solvent solid-state properties and the solubility in water of
evaporation) and operation parameters (fine didanosine re-crystallized from a dimethylsulf-
tuning of the density and the solvent power of oxide (DMSO) solution using supercritical CO2
the crystallization medium by modulating tem- as an antisolvent (SAS process) for comparison
perature and pressure) that cannot be obtained with the commercially available drug product.
with traditional crystallization processes from We also applied modern solid-state NMR (SS
liquid solvents, because of the obvious operational NMR) techniques, namely 2D 1H DQ CRAMPS
constraints (e.g., temperature range, cooling rate, (Combined Rotation And Multiple Pulse Spectro-
solute concentration variation due to the progres- scopy) and 1H–13C on- and off-resonance CP
sive separation of solid phases). (cross polarization) FSLG-HETCOR experiments,
Great attention has been recently devoted to known for providing reliable information about
1
drug re-crystallization by means of processes that H–1H21,22 and 1H–13C23–25 intra- and intermole-
use supercritical CO2 as an antisolvent (SAS, or cular proximities, in order to address polymorph-
Supercritical AntiSolvent).10–13 ism issues arising from the crystallization of a new

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010 DOI 10.1002/jps
DIDANOSINE POLYMORPHISM IN A SUPERCRITICAL ANTISOLVENT PROCESS 1857

form in the supercritical solvent process. These


pulse sequences allowed comparisons and con-
siderations about the crystal packing and hydro-
gen bonds since X-ray structures are not available
because of the unsuitable size of DDI specimens
for single crystal diffraction.26

EXPERIMENTAL

Chemicals
DDI was a kind gift of Bristol-Myers Squibb
Pharmaceutical Research Institute; CO2 (99.9%
grade, Sapio, Italy); DMSO, methanol lecithin,
cyclohexane ammonium acetate and concentrated
ammonia were of analytical grade. Figure 2. Schematic representation of the apparatus
used for SAS experiments: 1) CO2 reservoir; 2) High
Solubility of DDI in SC CO2 pressure pump; 3) Chilled bath; 4) CO2 inlet valve; 5) T
connector; 6) Precipitation vessel; 7) Three ways valve;
The solubility measurements of DDI in super- 8) Temperature controller; 9) Outlet valve; 10) Chilled
critical CO2 were performed by means of a bath; 11) DDI in DMSO solution reservoir; 12) HPLC
laboratory scale apparatus (SPE-ED SFE, Applied pump; 13) Methanol reservoir; 14) HPLC pump.
Separation, Allentown, PA), operating under
dynamic conditions at low flux according to the 4.5 mmol min1. The CO2 flux was manually
method described by Bettini et al.27 Experiments set by means of a micrometric valve positioned
were carried out at 40–458C and 90–100 bar. at the end of the pressurized circuit. After
15 min equilibration 1 mL of the DDI solution
(100 mg mL1) was pumped into the precipitation
SAS Re-Crystallization
vessel through the T valve. After complete DDI
The apparatus for the SAS crystallization was set precipitation, the vessel was submitted to a 2 h
up by modifying the apparatus used for solubility final washing with supercritical CO2 alone to
measurement (Fig. 2). Here the saturation cell ensure complete solvent removal from the pre-
was substituted with a precipitation vessel. cipitated solid particles and to prevent DMSO
In a typical SAS experiment the precipitation condensation during depressurization.
vessel (32 cm3 i.v., 20 cm length, 1.42 cm i.d.) was
connected through a low volume T valve with the
Thermal Analysis
lines of the pressurized CO2 and the DDI solution
in DMSO. The drug solution was pumped at Differential scanning calorimetry was performed
0.01 mL min1 by an HPLC pump (LKB-2248, on an indium calibrated Mettler DSC 821e
Pharmacia, Uppsala, Sweden). A sintered glass instrument (Mettler Toledo, Columbus, OH)
frit was positioned at the bottom of the precipita- driven by STARe software (Mettler Toledo).
tion vessel to collect the re-crystallized DDI. The DSC traces were recorded by placing accurately
temperature of the thermostatic chamber was set weighed amounts (5–6 mg) of powder sample in a
at 458C while the CO2 pressure was 100, 150, or 40 mL aluminum pan which was sealed and doubly
200 bar. pierced. Scans were performed between 25 and
A second HPLC pump (Waters-510 Millipore, 2008C at 10 K min1 under a flux of dry nitrogen
Milford, MA) pumped methanol (1 mL min1) (100 mL min1). Each powder sample was ana-
through the tubing below the precipitation vessel lyzed at least in triplicate.
in order to avoid possible obstruction stemming
from downstream precipitation of the drug. The
Powder X-Ray Diffractometry
circuit ended with a chilled bath for solvent
condensation. X-ray diffraction patterns on powder were record-
The CO2 was made to flow through the ed on a Miniflex (Rigaku, Tokyo, Japan) diffract-
precipitation vessel at a flow rate of about ometer using Cu Ka radiation (l ¼ 1.5418 Å)

DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
1858 BETTINI ET AL.

generated with 30 kV. The goniometer was set at a pulse length for protons was set at 2.0 ms while
scan rate of 0.58 min1 over the 2u interval 2–358. recycle delays used were 10–15 s. Quadrature
detection was achieved by using the states-TPPI
method. All the data for 40t1 increments with
Elemental Analysis 240 scans were collected. For the off-resonance CP
(LG-CP) FSLG-HETCOR NMR the intensity of
Elemental analysis for C, H and N was carried out the B1(1H) field for the CP was 75 kHz with a
with a FlashEA 1112 (ThermoQuest, Milan, Italy) mixing period of 2.0 ms. The 1H chemical shift
running at 9508C on powder samples of about 1 mg scale in the HETCOR spectra was corrected by a
accurately weighed in Sn crucibles. The accuracy scaling factor of 1/H3 since the 1H chemical-shift
of the analysis ranges between 0.01% and 0.3% of dispersion is scaled by a factor of 1/H3 during
the absolute value of element composition. FSLG decoupling.
1D 1H CRAMPS and 2D 1H DQ CRAMPS
experiments were performed with a 2.5 mm
Particle Size Analysis Bruker probe. 1H CRAMPS spectra were acquired
Particle size analysis was done with a Malvern using a windowed-PMLG pulse sequence of
Mastersizer X (Malvern Instruments LTD, dipolar decoupling at the spinning speed of
Malvern, UK) laser light diffractometer. Lenses 13 kHz. 2D 1H–1H DQ CRAMPS were acquired
of 45 and 100 mm ranges were used according to at 12.5 kHz of spinning speed with the window-
the sample needs and the laser beam length was less-eDUMBO-131 and the windowed-PMLG32
set at 2.40 mm. The samples were suspended in a pulse sequences for homonuclear dipolar decou-
1% solution of lecithin in cyclohexane, mixed in an pling during t1 and t2, respectively. For all
ultrasound bath and introduced into the light samples, 1H 908 pulse lengths of 1.9 ms and recycle
scattering cell by a Malvern stirring pump. delays of 10–15 s were used. For each of 64
increments of t1, 96 transients were averaged. The
pulse width and the RF power were finely
adjusted for best resolution and a minimal zero-
NMR Experiments
frequency line in the spectra. The q1 and q2
Solid-state NMR measurements were run on a prepulses were both of duration 0.5 ms (commer-
Bruker AVANCE II 400 instrument operating at cial) or 0.8 ms (SAS 200 bar). In t2, one complex
400.23, 100.65, and 40.55 MHz for 1H, 13C, and data point was acquired in each acquisition
15
N, respectively. 13C, 15N spectra were recorded window (2.2 ms). Including the prepulses, the
at room temperature at the spinning speed of effective dwell time in t2 was 16.1 ms (maximum
12 kHz. Cylindrical 4 mm o.d. zirconia rotors with t2 ¼ 11 ms). The DUMBO blocks were 25.6 ms long
sample volume of 120 mL were employed. For (corresponding to one basic DUMBO cycle). DQ
CPMAS experiments, a ramp cross-polarization excitation and reconversion was achieved using
pulse sequence was used with contact times of two elements of POST-C7,33 corresponding to a
3.5 ms (13C) or 4 ms (15N), a 1H 908 pulse of 3.30 ms, recoupling time of 45.7 ms. A 16-step nested phase
recycle delays of 10–15 s, and 128 (13C) or about cycle was used to select Dp ¼ 2 on the DQ
1000 (15N) transients. The two pulse phase- excitation pulses (four steps), and Dp ¼ 1 on the
modulation (TPPM) decoupling scheme was used z-filter 908 pulse (four steps). The States-TPPI
with a frequency field of 75 kHz. method was used to achieve sign discrimination in
For the spectral editing experiment a CPPISPI the F1 dimension.
sequence28 was used with a polarization-inversion 1
H, 13C, and 15N scales were calibrated
period of 50 ms. with adamantate (1H signal at 1.87 ppm), glycine
2D 13C–1H HETCOR spectra were measured (13C methylene signal at 43.86 ppm) and
according to the method of van Rossum et al.29,30 (NH4)2SO4 (15N signal at 355.8 ppm with respect
The MAS rate was set to 12 kHz. The proton rf to CH3NO2) as external standards.
field strength used during the t1 delay for FSLG
decoupling and during the acquisition for TPPM
Optical Microscopy
decoupling was 82 kHz. Two off-resonance pulses
with opposite phases (i.e., þx, x or þy, y) during Optical microscopy with polarized light was per-
the FSLG decoupling were set to 9.96 ms. The formed with an Optiphot2-Pol microscope (Nikon,
contact time was 100 ms. The magic angle (54.78) Tokyo, Japan) at 10 and 20 magnification.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010 DOI 10.1002/jps
DIDANOSINE POLYMORPHISM IN A SUPERCRITICAL ANTISOLVENT PROCESS 1859

DDI Solubility in Water phase B (30 volumes of methanol and 70 volumes


of a 3.86 g L1 solution of ammonium acetate
Weighed amounts of DDI (50 mg), in excess
adjusted to pH 8.0 with concentrated ammonia).
with respect to saturation, were added to 1 mL of
double distilled water and magnetically stirred in
EppendorfTM vials at 25  0.58C for 48 h. There-
after, DDI concentration in solution was deter- Powder Milling
mined by HPLC after filtering the suspensions
Weighed amounts (about 300 mg) of either com-
(0.45 mm, regenerated cellulose) and diluting the
mercial or re-crystallized DDI were ground for 2 or
filtrate with a solvent mixture constituted by
4 h, in a vibration mill (Mk 11, RIIC, Glenrothes,
8 volumes of mobile phase B and 92 volumes
Fife, Scotland, UK) equipped with an agate
of mobile phase A (see below HPLC analysis).34
cylindrical cell (33 mm height, 9.5 mm i.d.) con-
Each measurement was replicated at least three
taining 2 agate balls (6.0 mm diameter).
times.

RESULTS AND DISCUSSION


DDI Dissolution Rate
Drug dissolution experiments were carried out at The selection of the crystallization method as well
258C on accurately weighed amount of DDI as the operating condition in terms of temperature
(50 mg) using a USP 32 flow through dissolution and pressure range was based on preliminary
apparatus (Dissotest CE6, Sotax, Basel, CH). solubility measurements and literature data
Double distilled water flowing at 5 mL min1 relevant to CO2–DMSO phase behavior.35,36
was used as dissolution medium. Aliquots corre- Preliminary solubility measurements of DDI in
sponding to 1 mL of the flowing fluid were pure SC CO2 indicated the unsuitability of a
collected at the cell exit at fixed time (within process based on the rapid expansion of a super-
30 min) filtered (0.45 mm, regenerated cellulose) critical solution for the drug re-crystallization. In
and analyzed by HPLC for DDI quantification. fact, the DDI solubility values in supercritical
CO2, that is, at 40–458C and 90–100 bar, resulted
in the range 1  109–2  107 mole fraction,
values that are too low to afford a suitable yield.
High Performance Liquid Chromatography (HPLC)
On the other hand, DDI is very soluble in
DDI quantification was carried out according to DMSO34 which is among the most frequently used
the method reported in the European Pharmaco- organic solvents in the SAS process, owing to its
poeia34 using a high performance liquid chroma- good miscibility with SC CO2. The temperature
tographer constituted by 2 LC-10 ATvp pump, and and the pressure range used in the present work
a SPD 10Avp detector set at 254 nm (Shimadzu were based on the work of Gonzalez et al.35 and
Tokyo Japan) equipped with a Sperisorb1 5 mm Kordikowski et al.36 who reported the tempera-
ODS2 column (Waters Corporation, Milford, MA). ture and pressure conditions for complete mis-
The volume of injection was 20 mL (Rheodyne cibility of DMSO and CO2 in the supercritical
injector 7725i). The mobile phase was constituted phase.
by different volume ratios of mobile phase A SAS re-crystallization was carried out at 458C
(8 volumes of methanol and 92 volumes of a with a CO2 pressure ranging from 100 to 200 bar.
3.86 g L1 solution of ammonium acetate adjusted The yield of the re-crystallization process ranged
to pH 8.0 with concentrated ammonia) and mobile between 22% and 26%.

Table 1. Particle Size Distribution of Commercial DDI and DDI Re-Crystallized at Three Different Antisolvent
Pressures

Powder Dv, 0.1 (mm) Dv, 0.5 (mm) Dv, 0.9 (mm) Span
Commercial DDI 1.86 4.98 20.35 3.71
SAS DDI obtained at 100 bar 1.99 6.30 30.05 4.45
SAS DDI obtained at 150 bar 1.25 5.68 17.16 2.80
SAS DDI obtained at 200 bar 1.01 3.66 11.25 2.79

DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
1860 BETTINI ET AL.

dynamics: the pressure increase implies an


increase of the fluid density and velocity and,
therefore, of the Reynolds number which, in turn,
results in a better mixing efficiency between the
organic solution and the supercritical fluid.6
Another parameter that should be taken into
account is the supercritical fluid to drug solution
volume ratio. The fluxes of the organic solution of
the drug and the flux of the CO2, measured after
the final expansion valve (at ambient conditions),
were kept constant at 0.01 and 100 mL min1,
respectively. The density values of CO2 at given
pressures and temperatures were calculated
by means of the Refprop software (NIST,
Figure 3. Cumulative undersize distribution of DDI
particles obtained by SAS re-crystallization at three
Gaithersburg, MD). Based on these values, the
different antisolvent pressures. ratios of the volume of the organic solution and the
supercritical CO2 into the precipitation vessel
were calculated as 39.7, 26.6, and 24.3 at 100, 150,
The particle size distribution of the SAS and 200 bar, respectively. The progressive reduc-
re-crystallized DDI is reported in Table 1 as a tion of this ratio with the pressure (and therefore
function of the CO2 pressure in comparison with with the density) resulted in an increase of the
the commercially available drug. concentration of the drug solution at the super-
The particle size and the particle size distribu- saturation point when the antisolvent effect took
tion (the Span value is defined as the ratio place.37
Dv, 0.9  Dv, 0.1/Dv, 0.5) of the powder obtained at Beside the particle size, also the crystal
100 bar were higher compared to the commercial morphology was affected by the SAS process.
DDI powder. On the other hand, for the re- Commercial DDI (Fig. 5A) appeared as aggregates
crystallized powders, a significant and progres- of irregular shaped crystals, while SAS re-crystal-
sive decrease of the particle size and particle size lization afforded needle shaped and more regular
distribution can be observed with the increased crystals (Fig. 5B and C). No effect of pressure can
pressure (Fig. 3). In agreement with the findings be evidenced on the crystal habit of the DDI re-
of other authors,13 the geometric mean diameter crystallized under different conditions.
of the SAS processed particles decreases with the As far as the purity of the crystallized compound
increase of the CO2 density (Fig. 4). is concerned, in particular with respect to possible
The phenomenon can be mainly ascribed to an traces of residual DMSO, elemental analysis was
effect of the CO2 pressure variation on the fluid carried out on the SAS re-crystallized DDI as well
as on the commercially available product. Accord-
ing to the pharmacopoeial classification DMSO
belongs to Class 3 which comprises the solvents of
low toxic potential and whose daily exposure
should not exceed 50 mg.38 In this respect, for a
DDI daily dose of 400 mg39 the maximum accep-
table DMSO concentration in the active ingredi-
ent would be 12.5% (w/w). Nevertheless, the
analysis carried out on DDI samples re-crystal-
lized by SAS at 100 and 150 bar (Tab. 2) indicated
a C, H, and N concentration very close to the
values measured for the commercial product and
practically superimposed to the theoretical ones
calculated on the base of the C10H12N4O3 formula.
Table 2 reports also the C, H, and N percentages
Figure 4. Geometric mean diameter of DDI particles relevant to the hypothetical case of a DMSO
obtained by SAS re-crystallization as a function of CO2 content of 0.5%, which should be considered as a
density. very low value in relation to the toxicity of the

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010 DOI 10.1002/jps
DIDANOSINE POLYMORPHISM IN A SUPERCRITICAL ANTISOLVENT PROCESS 1861

Many literature reports underscored the effect


of the supercritical process on the solid-state
characteristics of the treated drug in particular
with respect to the degree of crystallinity, crystal
phase and polymorph generation.11,13,40,41
DSC analysis carried out on commercial
DDI evidenced melting with decomposition at
186  0.18C. Traces that displayed no significant
differences were obtained from the re-crystallized
powders, that showed melting point temperatures
of about 18C around the melting peak temperature
of the commercial DDI. However, these variations
could not be correlated to the experimental con-
ditions (pressure value or CO2 density) adopted
during the SAS process. The composite nature of
the phenomenon (endothermic melting with con-
comitant exothermic decomposition) makes the
DSC unsuitable for detecting possible melting
point variations stemming from different crystal
phases.
In this respect powder X-ray diffraction proved
to be a more appropriate tool.
Figure 6 reports the diffraction patterns
obtained from commercial and SAS re-crystallized
(200 bar) DDI. Nonsuperimposable traces were
obtained. Main differences are in the 8–118,
12–138, 16–208, and 26–288 2u regions. This
observation strongly suggested that a new poly-
morph was isolated upon SAS re-crystallization.
To our best knowledge, no data relevant to
DDI polymorphism are presently available in
literature nor on the Cambridge Structural
Database (http://www.ccdc.cam.ac.uk/products/
csd/).
The powders obtained by SAS at different
Figure 5. Pictures taken at the polarized light pressures did not show any difference in PXRD
microscope of (A) commercial DDI crystals and SAS patterns, thus suggesting that the formation of
re-crystallized DDI B) 100 bar, C) 200 bar. Original a new crystal phase might be ascribed to a
magnification 20X. combination of both the organic solvent charac-
teristics and the crystallization conditions im-
solvent. It is noteworthy that the relevant figures posed by the supercritical milieu.
are far away from the measured one. This allows In fact, considering that DMSO shows quite a
one to conclude that the re-crystallized products low vapor tension (boiling point 1898C at ambient
present a negligible residual solvent content. pressure) the supercritical fluid played a crucial

Table 2. Composition Percent of the Commercial DDI and DDI Obtained after SAS Re-Crystallization at 100 and
150 bar along With the Theoretical Values Calculated for Pure DDI or DDI Containing 0.5% (w/w) of DMSO

Theoretical Measured

Element % Pure DDI DDI þ 0.5% DMSO SAS DDI (100 bar) SAS DDI (150 bar) Commercial DDI
C 50.84 44.79 50.21 50.04 50.10
H 5.12 4.54 5.16 5.14 5.08
N 23.71 20.81 23.72 23.77 23.91

DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
1862 BETTINI ET AL.

Figure 6. X-Ray diffraction patterns of commercial DDI powder and SAS re-crystal-
lized DDI (200 bar) powder.

role in the crystal formation in a very short mities.51,52 Furthermore, unambiguous 1H signal
time. In this respect, the otherwise difficult to assignments can be achieved by on-resonance CP
1
realize rapid solvent evaporation determined by H–13C HETCOR in which polarization transfer
the antisolvent effect exerted by the SC CO2, made occurs during a very short time, namely the
affordable the crystallization process. contact time (CT). For longer CT, the detection of
additional 1H–13C coherences provides further
geometrical constraints if analyzed with respect to
Solid-State NMR Analysis the formation of 1H–1H polarization coherences.25
For a complete analysis, the use of an off-
It is well known that advanced solid-state NMR resonance CP achieved by LG spin-lock (LG–
techniques provide several structural and CP) efficiently suppresses unwanted 1H–1H spin
dynamic data useful in many areas, in particular exchange leading to the observation of long-range
1
those concerning polymorphism in pharma- H–13C contacts.29
ceutical,4,42 organic,43–45 and in crystal engineer- For these reasons the solid-state characteristics
ing46,47 fields. Due to recently developed of the SAS crystallized DDI in comparison to those
techniques based on the dipole–dipole interaction of the commercially available product were
between homo- and heteronuclei,23 solid-state further investigated by 1H CRAMPS, 13C, and
15
NMR is becoming essential where X-ray struc- N CPMAS, spectral editing 13C CPPISPI (Cross
tures are not available.48,49 These experiments Polarization Polarization Inversion Simultaneous
are nowadays able to give sufficient structural Phase Inversion), 1H DQ CRAMPS and 1H–13C
information to elucidate conformation and aggre- on- and off-resonance CP (LG-CP) FSLG-HET-
gation even for unlabelled systems.23 Recent COR NMR.
1
improvements in probe technology with spinning H and 13C chemical shifts with assignments
speed up to 70 kHz and in homodecoupling are listed in Table 3 while 15N chemical shifts are
techniques employing the Lee–Goldburg (LG) reported in Table 4.
approach (e.g., FSLG, frequency switched, or The 13C solution spectrum of the DDl53 shows
PMLG, phase modulated)32,50 allow the acquisi- many differences with respect to 13C CPMAS
tion of high resolution 1H spectra in both 1D and spectra of commercial and SAS crystallized DDI
2D modes. Several CRAMPS schemes have been (Fig. 7) related to significant conformational
used in order to elucidate hydrogen bond net- changes and/or geometrical constraints due to
works in many systems while the DQ spectra crystallization forces such as weak interactions or
provide reliable information about 1H–1H proxi- hydrogen bonds.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010 DOI 10.1002/jps
DIDANOSINE POLYMORPHISM IN A SUPERCRITICAL ANTISOLVENT PROCESS 1863

1 13
Table 3. H and C Chemical Shifts With Assignment for the Commercial and SAS 200 bar DDI Samples

Solution-State NMRa Solid-State NMR

Commercial, Commercial, 200 bar, 200 bar,


1 13 1 13 1 13
Atom Note H (ppm) C (ppm) H (ppm) C (ppm) H (ppm) C (ppm)
C6 C–
–O — 157.5 156.9 156.6
155.4 155.2
C2 (H)NC(H)N 8.34 (s, 1H) 146.5 8.0 149.3 (2) 8.5 148.3 (2)
C4 C(q) — 148.4 147.8 148.3 (1)
145.5 145.5
C8 (R)NC(H)N 8.05 (s, 1H) 139.1 7.7 138.5 8.0 137.3
7.3 136.0 7.6 135.3
C5 C(q) — 125.2 123.8 123.7
122.0 121.9
C10 , C40 CH 6.20 (dd, 1H) 85.3 4.7 85.3 (2) 5.7 85.0 (3)
4.10–4.13(m, 2H) 82.9 5.3 84.7 (2) 5.1 83.9
C50 CH2OH 3.53–3.62(m, 2H) 63.5 3.8 61.7 (2) 4.2 61.7 (2)
C20 CH2 2.35–2.51(m, 2H) 33.0 2.3 33.9 2.6 33.2
2.3 30.9 3.0 30.2
C30 CH2 1.99–2.06 (m, 2H) 26.3 2.3 26.5 2.5 25.6
2.4 23.7 2.7 24.4
NH 12.33 (brs, 1H) 10.4/14.7 10.3/14.9
OH 5.0 (brs, 1H) 14.7/10.4 14.9/10.3
a
From Ref.,53 DMSO-d6 solvent.

The solid-state 13C assignment was obtained


15
Table 4. N Chemical Shifts With Assignment for with the help of the CPPISPI experiment (Fig. 8)
the Commercial and SAS 200 bar DDI Samples that allows the discrimination of carbon atoms
according to the number of attached protons based
Atom Note Commercial 200 bar on differences in CP dynamics. We optimized the
N(1) NH 152.8 153.6 polarization inverse period to 50 ms in order to
N(1) NH 146.3 146.0 obtain positive C and CH3 signals, negative CH2
N(3) N(q) 186.7 186.9 and suppressed CH peaks.
N(3) N(q) 178.9 179.5 Almost all carbon atoms of commercial DDI
N(7) N(q) 211.3 211.1 gave rise to two peaks, due to the presence of two
N(9) N(q) 159.5 159.7
N(9) N(q) 156.6 156.4

Figure 8. 13C CPMAS (A) and 13C CPPISPI (spectral


editing) (B) spectra of the commercial sample of the
DDI. In the CPPISPI spectrum the polarization-inver-
Figure 7. 13C CPMAS spectra of the commercial and sion period was set to 50 ms in order to obtain
SAS 100, 150, and 200 bar DDI recorded at 100.63 MHz positive Cq, negative CH2 and null CH (marked with
with a spinning speed of 12 kHz. circles).

DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
1864 BETTINI ET AL.

Figure 9. 1H CRAMPS spectra of the commercial and


Figure 10. Comparison among 15N CPMAS spectra
SAS 200 bar DDI, recorded at 400.23 MHz with a spin-
of commercial DDI and SAS re-crystallized didanosine
ning speed of 13 kHz.
at 100, 150, and 200 bar of CO2.

independent molecules in the crystallographic C5 would be strongly influenced, leaving all other
asymmetric unit. This aspect might also be a resonances almost unchanged, whereas this
consequence of a keto-enol tautomeric equili- was not observed. The signal C6 (C –– O) at
brium, but in this case signals C6, C2, C4, and 156.6 ppm might indicate that both molecules

Figure 11. 2D 1H–13C FSLG-HETCOR (A) and off-resonance CP 1H–13C FSLG-HET-


COR (B) spectrum of commercial DDI, recorded with a contact time of 100 and
2000 ms, respectively.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010 DOI 10.1002/jps
DIDANOSINE POLYMORPHISM IN A SUPERCRITICAL ANTISOLVENT PROCESS 1865

exist in the enol form: indeed, it has been reported simple acquisition of 1H MAS spectra at 35 kHz
for similar systems that the C – O group did not provide enough resolution, so we decided
resonates in the solid state at about 180 ppm, to use a CRAMPS approach by means of the
while C–OH groups give rise to peaks around windowed-PMLG scheme.32 The 1H CRAMPS
167 ppm,54 though they may shift to 167–166 and spectra of the commercial and SAS crystallized
154–155 ppm respectively, depending on ring DDI are reported in Figure 9. They are quite
substituents.55 similar and characterized by five resonances, two
As far as the SAS re-crystallized DDI was of which (at 10.5 and 14.8 ppm) are clearly related
concerned, spectra of powders obtained at 100, to hydrogen bonded atoms involved in a weak and
150, and 200 bar were identical (Fig. 7). Therefore, in a strong interaction, respectively.
only the sample prepared at 200 bar will be The 15N spectra are characterized by seven
discussed. signals (Fig. 10): the comparison of the com-
The 13C CPMAS spectrum of SAS re-crystal- mercial and the SAS crystallized samples shows a
lized DDI was quite similar to that of the shift in the range of 0.2–0.5 ppm for all reso-
commercial form. However, small but significant nances. Main differences involve (a) the N1 atom
differences confirmed the isolation of a new that shifts from 152.8 to 153.6 ppm with increas-
crystalline product. The main changes involve ing of the line width at half height (from 48 to
resonances C30 and C10 : the former in SAS re- 57 Hz) and (b) the sharpening of N7 atom from
crystallized samples gave rise to two signals at 100 to 40 Hz.
25.6 and 24.4 ppm, closer than those of the After the 13C assignment, 1H resonances were
commercial product. In SAS DDI the C10 signal detected from one-bond correlation signals high-
is split into two peaks (83.9 and 85.0 ppm), one of lighted by the on-resonance CP 1H–13C FSLG-
which is overlapped with C40 , while in the HETCOR spectrum (Fig. 11A) acquired with a
commercial sample C10 atoms gives rise to a
single peak (84.7 ppm). Furthermore, in the SAS
DDI, C8 resonances are shifted to lower frequen-
cies of about 1.2 and 0.7 ppm, respectively,
whereas the high frequency C4 peak is completely
overlapped with C2. The remaining signals did
not show any significant variations compared to
the commercial product.
By comparing solution chemical shifts (which
refer to monomeric or dissolved structures) with
solid-state chemical shifts (related to aggregated
or crystallized structures) we observed the highest
shifts (dliq–d) for carbon atoms C4, C8, and C5 with
a Dd of 2.9–2.9, 3.1–3.8, and 3.2–3.3 ppm, respec-
tively (reported values refer to the most shifted
peak only of the two independent molecules; first
value, commercial DDI, second value SAS DDI).
These carbon atoms experience the strongest
packing forces, probably because involved in
noncovalent bonding. In the case of C8, this is
in agreement with its ability to correlate
through the space with many other hydrogen
atoms (see below). The presence of a single signal
at 157.5 ppm for C6 in solution suggests the
existence of a tautomeric equilibrium probably
shifted to the enol form.
Figure 12. 2D 1H DQ CRAMPS spectrum of the com-
Usually, the easiest technique for averaging mercial DDI together with the single-quantum projec-
the strong 1H–1H homonuclear dipolar interac- tions. The pair of DQ signals in both spectra
tion present in organic solids is to spin the corresponding to the intramolecular proximity of hydro-
sample in small rotors (1.3–2.5 mm o.d.) up to gen-bonded protons with high-ppm resonance signals to
30–70 kHz.56,57 Unfortunately, in this case, the the same nearby proton are highlighted.

DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
1866 BETTINI ET AL.

very short CT (100 ms). Indeed, cross peaks among all proton and carbon atoms of the pyrrole
relating to long-range correlations or due to moiety. We were able to trace correlations from
1
H spin diffusion are almost completely sup- the proton H8 to carbon atoms C10 , C40 , C4, C5,
pressed. On the other hand, the off-resonance CP and C6. All these proximities, indicating a
1
H–13C FSLG-HETCOR experiment (Fig. 11B), particular conformation of the pyrrole moiety
acquired with a CT of 2000 ms, allowed una- around the C8–H8 group, explain its large shift
mbiguous assignments of long-range spatial with respect to the solution state. We also
1
H–13C correlations since the LG 1H spin-lock observed polarization transfers from H20 and
avoids 1H–1H spin exchange during the CP. In H30 to C8 and from H10 and H40 to C4. Further-
this way, structural information such as confor- more, both hydrogen bonded signals show correla-
mations, angles and distances becomes poten- tions with the carbon C6 confirming the presence
tially available allowing an NMR crystallography of an enol (C–OH) rather than a carbonyl (C – O)
analysis that in some case provides accurate group. Interestingly they transfer polarization
short-range structures.58,59 Unfortunately, the also to carbon atoms C2 and C5 in agreement
presence of two molecules in the asymmetric unit with the presence of N1  H–O or N3  H–O
dramatically complicates the analysis: indeed, interactions.
while two 13C signal sets are clearly visible (for These results suggest that the two molecules in
almost all carbon atoms), it is not the same for the asymmetric unit possess very similar con-
1
H resonances. However, even if it is difficult to formations, while the main difference stems in the
distinguish between inter- and intramolecular hydrogen bond strength (signals at 10.4 and
correlations, several conclusions can be drawn. 14.7 ppm which refer to the same C6–O–
All long-range correlations are highlighted in H1  N interaction in the two molecules both
Figure 11B. Polarization transfer is observed correlating with carbon atoms C6). They differ for

Figure 13. 2D 1H–13C FSLG-HETCOR (A) and off-resonance CP 1H–13C FSLG-HET-


COR (B) spectrum of SAS 200 bar DDI, recorded with a contact time of 100 and
2000 ms, respectively.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010 DOI 10.1002/jps
DIDANOSINE POLYMORPHISM IN A SUPERCRITICAL ANTISOLVENT PROCESS 1867

The solubility in water was determined at 258C


for investigating possible differences between
commercial and re-crystallized DDI, as well as,
as a coarse indicator of the relative thermody-
namic stability of the two crystal forms.
The values obtained from the commercial DDI
and the DDI SAS re-crystallized at 200 bar were
25.4  0.7 and 26.9  0.5 mg mL1, respectively.
The difference was very low but statistically
significant ( p < 0.05 by Student’s t-test) and
suggested that the small variation observed in
the crystal conformation played a role on the
energetic content of the molecule. Therefore it
could be affirmed that the SAS re-crystallized DDI
represents a metastable polymorph of the com-
mercial DDI.60,61
It is worth underscoring that the accurate
determination of equilibrium solubility of two
polymorphs is often a difficult task. In fact,
solution mediated conversion of the less stable
form to the more stable one may prevent dif-
ferences in solubility from being detected.
Nevertheless, this small difference in solubility
Figure 14. 2D 1H DQ CRAMPS spectrum of the SAS observed resulted in a difference, although not
200 bar DDI, together with the single-quantum projec- statistically significant, in dissolution rate bet-
tions. The pair of DQ signals in both spectra correspond- ween the re-crystallized and commercial DDI
ing to the intramolecular proximity of hydrogen-bonded (Fig. 15).
protons with high-ppm resonance signals to the same Finally, a further indication of the lower
nearby proton are highlighted. thermodynamic stability of the re-crystallized
DDI with respect to the commercial product
the correlation with C5 only present for the signal stemming from the different energy content of
at 14.7 ppm. the crystal lattice was obtained by submitting
Also the hydrogen bond network seems to be the two powders to prolonged milling using a
very similar, as observed in the 1H DQ CRAMPS vibration mill.
(Fig. 12) where both hydrogen-bonded atoms show The decrease of the intensity of the X-ray
proximity to H8. diffraction peaks indicated that the commercial
The same analysis has been performed on the product underwent a reduction of the degree
SAS DDI 200 bar (Figs. 13 and 14). Interestingly, of crystallinity after 2 h milling Figure 16A. No
the LG-CP HETCOR spectrum highlights the
same correlations observed in the commercial
sample, somewhat more intense. Furthermore,
by comparing 1H DQ CRAMPS spectra (Figs. 12
and 14), we can assert that hydrogen bond
correlation patterns for the two forms are equal
as well as all other hydrogen proximities, of course
by taking into account small differences in the
chemical shifts. Then, the two forms possess
very similar structural conformation but at
least differences in the molecular packing. If we
consider that also the interaction networks are
similar we can surmise that they probably differ
mainly in the cell volume and density, that is, in
the interaction lengths and in the intermolecular Figure 15. Dissolution profile of commercial DDI and
distances. DDI re-crystallized at 200 bar.

DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
1868 BETTINI ET AL.

diffractometry and well described by solid-state


NMR. The carbon C6 13C chemical shift suggests
that both DDI samples are in the enol form. The
analysis of homo- and heteronuclear proximities
obtained by means of 2D NMR experiments shows
that commercial and SAS 200 bar DDI possess
very similar conformation and hydrogen bond
network, but different packing.
The new polymorph proved to be a metastable
form showing higher solubility in water and lower
stability to mechanical stress.
The particle size of the new crystal phase can
be reduced by varying the antisolvent density
through a pressure increase.

ACKNOWLEDGMENTS

This work was supported by a grant from Italian


Ministry for University and Research through the
PRIN 2006 program. The authors would like to
thank Prof. Mino R. Caira (University of Cape
Town, South Africa) for the helpful discussion.

REFERENCES

1. Brittain HG. 2007. Polymorphism and solvato-


morphism 2005. J Pharm Sci 96:705–728.
2. Brittain HG. 2008. Polymorphism and solvato-
morphism 2006. J Pharm Sci 97:3611–3636.
3. Brittain HG. 2009. Polymorphism and solvato-
morphism 2007. J Pharm Sci 98:1617–1642.
4. Hilfiker R. 2006. Polymorphism in the pharmaceu-
Figure 16. Powder X-ray diffraction patterns of
tical industry. Weinheim: Wiley-VCH. pp. 649–650.
(A) commercial DDI and (B) SAS re-crystallized DDI
5. Blagden N, de Matas M, Gavan PT, York P. 2007.
at 200 bar recorded upon different milling time.
Crystal engineering of active pharmaceutical ingre-
dients to improve solubility and dissolution rates.
Adv Drug Del Rev 59:617–630.
further decrease was observed upon 4 h milling.
6. Pasquali I, Bettini R, Giordano F. 2008. Supercri-
On the contrary, the SAS re-crystallized powder tical fluid technologies: An innovative approach for
(200 bar) showed a progressive decrease of the manipulating the solid-state of pharmaceuticals.
peak intensity with the milling time, accompanied Adv Drug Del Rev 60:399–410.
by an elevation of the baseline, that indicated a 7. Pasquali I, Bettini R. 2008. Are pharmaceutics
reduction of the degree of crystallinity with the really going supercritical? Int J Pharm 364:176–
formation of an amorphous phase (Fig. 16B). 187.
8. York P. 1999. Strategies for particle design using
supercritical fluid technologies. Pharm Sci Technol
CONCLUSIONS 2:430–440.
9. Bettini R, Rossi A, Lavezzini E, Pasquali I, Gior-
dano F. 2003. Thermal and morphological charac-
A new polymorph of DDI was produced by a
terization of micronized acetylsalicylic acid powders
supercritical antisolvent process and charac- prepared by rapid expansion of a supercritical solu-
terized by means of 1D and 2D multinuclear tion. J Therm Anal Calorim 73:487–497.
(1H, 13C, 15N) SS NMR. The structural differences 10. Park SJ, Yeo SD. 2008. Recrystallization of caffeine
between the commercial product and the SAS re- using gas antisolvent process. J Supercrit Fluids 47:
crystallized DDI are underlined by powder X-ray 85–92.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010 DOI 10.1002/jps
DIDANOSINE POLYMORPHISM IN A SUPERCRITICAL ANTISOLVENT PROCESS 1869

11. Park EJ, Kim M-S, Lee S, Kim J-S, Woo J-S, Park 23. Brus J, Jegorov A. 2004. Through-bonds and
JS, Hwang S-J. 2007. Recrystallization of flucona- through-space solid-state NMR correlations at nat-
zole using the supercritical antisolvent (SAS) pro- ural isotopic abundance: Signal assignment and
cess. Int J Pharm 328:152–160. structural study of Simvastatin. J Phys Chem A
12. Meng D, Falconer J, Krauel-Goellner K, Chen 108:3955–3964.
JJJJ, Farid M, Alany RD. 2008. Self-built super- 24. van Rossum BJ, Förster H, de Groot HJM. 1997.
critical CO2 anti-solvent unit design, construc- High-field and high-speed CP-MAS13C NMR
tion and operation using carbamazepine. AAPS heteronuclear dipolar-correlation spectroscopy of
PharmSciTech 9:944–952. solids with frequency-switched Lee–Goldburg
13. Kim M-S, Jin S-J, Kim J-S, Park HJ, Song H-S, homonuclear decoupling. J Magn Reson 124:516–
Neubert RHH, Hwang S-J. 2008. Preparation, char- 519.
acterization and in vivo evaluation of amorphous 25. Chierotti MR, Gobetto R, Pellegrino L, Milone L,
atorvastatin calcium nanoparticles using supercri- Venturello P. 2008. Mechanically induced phase
tical antisolvent (SAS) process. Eur J Pharm Bio- change in barbituric acid. Cryst Growth Des 8:
pharm 69:454–465. 1454–1457.
14. Lalanne M, Andrieux K, Paci A, Besnard M, Ré M, 26. Caira MR. 2009. Personal communication, to
Bourgaux C, Ollivon M, Desmaele D, Couvreur P. Bettini R, Parma.
2007. Liposomal formulation of a glycerolipidic pro- 27. Bettini R, Bonassi L, Castoro V, Rossi A, Zema L,
drug for lymphatic delivery of didanosine via oral Gazzaniga A, Giordano F. 2001. Solubility and
route. Int J Pharm 344:62–70. conversion of carbamazepine polymorphs in super-
15. Jain SK, Gupta Y, Jain AS, Saxena AR, Khare P, critical carbon dioxide. Eur J Pharm Sci 13:281–
Jain A. 2008. Mannosylated gelatin nanoparticles 286.
bearing an anti-HIV drug didanosine for site-spe- 28. Wu XL, Zilm KW. 1993. Complete spectral editing
cific delivery. Nanomed: Nanotechnol Biol Med 4: in CPMAS NMR. J Magn Reson 102:205–
41–48. 213.
16. Jin Y, Ai P, Xin R, Chen D. 2008. Morpholo- 29. van Rossum BJ, de Groot CP, Ladizhansky V, Vega
gical transformation of self-assembled nanostruc- S, de Groot HJM. 2000. A method for measuring
tures prepared from cholesteryl acyl didanosine heteronuclear (1H–13C) distances in high speed
and the optimal formulation of nanoparticulate MAS NMR. J Am Chem Soc 122:3465.
systems: Effects of solvents, acyl chain length 30. van Rossum BJ, Schulten EAM, Raap J, Oschkinat
and poloxamer 188. J Coll Interf Sci 326:275– H, deGroot HJM. 2002. A 3-D structural model of
282. solid self-assembled chlorophyll a/H2O from multi-
17. Jin Y, Xin R, Ai P, Chen D. 2008. Self-assembled spin labeling and MAS NMR 2-D dipolar correlation
drug delivery systems 2. Cholesteryl derivatives of spectroscopy in high magnetic field. J Magn Reson
antiviral nucleoside analogues: Synthesis, proper- 155:1–1-14.
ties and the vesicle formation. Int J Pharm 350: 31. Elena B, de Paëpe G, Emsley L. 2004. Direct spec-
330–337. tral optimisation of proton–proton homonuclear
18. Kim DD, Chien YW. 1996. Transdermal delivery dipolar decoupling in solid-state NMR. Chem Phys
of dideoxynucleoside-type anti-HIV drugs. 2. The Lett 398:532–538.
effect of vehicle and enhancer on skin permeation. 32. Vinogradov E, Madhu PK, Vega S. 1999. High-
J Pharm Sci 85:214–219. resolution proton solid-state NMR spectroscopy
19. Mukherji E, Millenbaugh N, Au J. 1994. Percuta- by phase-modulated Lee–Goldburg experiment.
neous absorption of 20 ,30 -dideoxynosine in rats. Chem Phys Lett 314:443–450.
Pharm Res 11:809–815. 33. Hohwy M, Jakobsen HJ, Edén M, Levitt MH,
20. Ojewole E, Mackraj I, Naidoo P, Govender T. 2008. Nielsen NC. 1998. Broadband dipolar recoupling
Exploring the use of novel drug delivery systems for in the nuclear magnetic resonance of rotating
antiretroviral drugs. Eur J Pharm Biopharm 70: solids: A compensated C7 pulse sequence. J Chem
697–710. Phys 108:2686.
21. Brown SP, Lesage A, Elena B, Emsley L. 2004. 34. European Pharmacopoeia. 2007. Ph.Eur. mono-
Probing proton-proton proximities in the solid graph 2200. 6th edition. European Pharmacopoeia.
state: high-resolution two-dimensional 1H–1H dou- EDQM, Strasbourg, France, p. 635.
ble-quantum CRAMPS NMR spectroscopy. J Am 35. Gonzalez AV, Tufeu R, Subra P. 2002. High-
Chem Soc 126:13230–13231. pressure vapor-liquid equilibrium for the binary
22. Griffin JM, Martin DR, Brown SP. 2007. Distin- systems carbon dioxide þ dimethyl sulfoxide and
guishing anhydrous and hydrous forms of an active carbon dioxide þ dichloromethane. J Chem Eng
pharmaceutical ingredient in a tablet formulation Data 47:492–4495.
using solid-state NMR spectroscopy. Angew Chem 36. Kordikowski A, Schenk AP, Van Nielen RM, Peters
Int 46:8036–8038. CJ. 1995. Volume expansion and vapor-liquid

DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
1870 BETTINI ET AL.

equilibria of binary mixtures of a variety of polar zeolite crystal structures. J Am Chem Soc 127:
solvents and certain near-critical solvents. 10365–10370.
J Supercrit Fluids 8:205–216. 49. Elena B, Emsley L. 2005. Powder crystallography
37. Weber A, Kummel R, Kraska T. 2004. Investigation by proton solid-state NMR spectroscopy. J Am
and modelling of the gas-antisolvent process. In: Chem Soc 127:9140–9146.
Brunner G, editor. Supercritical fluids as solvents 50. Bielecki A, Kolbert AC, Levitt M. 1989. Frequency-
and reaction media. Amsterdam: Elsevier. pp. 429– switched pulse sequences: Homonuclear decoupling
448. and dilute spin NMR in solids. Chem Phys Lett 185:
38. European Pharmacopoeia. 2008. 5.4 Residual 341–346.
Solvents. European Pharmacopoeia EDQM, 51. Brown SP, Spiess HW. 2001. Advanced solid-
Strasbourg, France. state NMR methods for the elucidation of structure
39. Medical Economics Company, Inc. 2002. Physi- and dynamics of molecular, macromolecular, and
cian’s desk reference. Medical Economics Company, supramolecular systems. Chem Rev 101:4125–
Inc. p. 1144. 4155.
40. Bettini R, Zampieri M, Martini A, Fumagalli P, 52. Chierotti MR, Gobetto R. 2008. Solid-state NMR
Rossi A, Giordano F. 2002. AAPS Annual Meeting, studies of weak interactions in supramolecular sys-
p. T2328. tems. Chem Commun 1621–1634.
41. Kordikowski A, Shekunov T, York P. 2001. Poly- 53. Srinivasa Rao DVN, Srinivas N, Bharathi C, Pra-
morph control of sulfathiazole in supercritical CO2. sad CS, Dandala R, Naidu A. 2007. Identification
Pharm Res 18:682–688. and characterization of new impurity in didanosine.
42. Harris RK. 2007. Applications of solid-state NMR to J Pharm Biomed Anal 45:516–520.
pharmaceutical polymorphism and related matters. 54. Harris RK, Olivieri AC. 1992. Quadrupolar effects
J Pharm Pharmacol 59:225–239. transferred to spin- magic-angle spinning spectra
43. Bechinger B, Skladnev DA, Ogrel A, Li X, Rogozh- of solids. Progr Nucl Magn Res Spectr 24:435–
kina EV, Ovchinnikova TV, O’Neil JDJ, Raap J. 456.
2001. 15N and 31P solid-state NMR investigations 55. Pizzala H, Carles M, Stone WEE, Thevand A. 2000.
on the orientation of zervamicin II and alamethicin Tautomerism in Schiff bases derived from 3-hydro-
in phosphatidylcholine membranes. Biochemistry xysalicylaldehyde. Combined X-ray diffraction,
40:9428–9437. solution and solid state NMR study. J Chem Soc
44. Braga D, Maini L, Fagnano C, Taddei P, Chierotti Perkin Trans 2:935–939.
MR, Gobetto R. 2007. Polymorphism in crystalline 56. Hafner S, Demco DE. 2002. Solid-state NMR spec-
cinchomeronic acid. Chem A Eur J 13:1222–1230. troscopy under periodic modulation by fast magic-
45. Braga D, Grepioni F, Maini L, Polito M, Rubini K, angle sample spinning and pulses: A review. Solid
Chierotti MR, Gobetto R. 2009. Hetero-seeding and State Nucl Magn Reson 22:247–274.
solid mixture to obtain new crystalline forms. Chem 57. Traer JW, Montoneri E, Samoson A, Past J, Tuher
A Eur J 15:1508–1515. JT, Goward GR. 2006. Unraveling the complex
46. Braga D, Grepioni F, Chierotti MR, Gobetto R. hydrogen bonding of a dual-functionality proton
2008. Three polymorphic forms of the co-crystal conductor using ultrafast magic angle spinning
4,4-bipyridine/pimelic acid and their structural, NMR. Chem Mater 18:4747–4754.
thermal, and spectroscopic characterization. Chem 58. Harris RK. 2004. NMR crystallography: The use
Eur J 14:10149–10159. of chemical shifts. Solid State Sci 6:1025–
47. Braga D, Grepioni F, Polito M, Chierotti MR, Ellena 1037.
S, Gobetto R. 2006. A solid-gas route to polymorph 59. Harris RK. 2006. NMR studies of organic poly-
conversion in crystalline [FeII(h5-C5H4COOH)2]. morphs and solvates. Analysis 131:351–373.
A diffraction and solid-state NMR study. Organo- 60. Carstensen JT. 2001. Advanced pharmaceutical
metallics 25:4627–4633. solids. Madison, Wisconsin: Marcel Dekker, Inc.
48. Brouwer DH, Darton RJ, Morris RE, Levitt MH. 61. Brittain HG. 1999. Polymorphism in pharmaceuti-
2005. A solid-state NMR method for solution of cal solids. New York: Marcel Dekker, Inc.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010 DOI 10.1002/jps

Anda mungkin juga menyukai