Physical Stability Studies of Miscible Amorphous Solid
Dispersions
IGOR IVANISEVIC
SSCI, A Division of Aptuit, W. Lafayette, Indiana 47906
Received 26 January 2010; revised 26 March 2010; accepted 26 April 2010
Published online 8 June 2010 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.22247
ABSTRACT: Physical stability of 12 amorphous solid dispersions was evaluated over 9 22
months under ambient conditions using X-ray powder diffraction. The nine dispersions initially
characterized as miscible drug polymer systems all remained X-ray amorphous for the duration
of their respective studies. In contrast, the three phase-separated systems all crystallized in 1 2
months, while the pure amorphous active pharmaceutical ingredients used in this study all
crystallized within a few days, under the conditions of this study. Changes in the local order of
dispersions that included polyvinylpyrrolidone were observed and appeared to correlate to
periods of higher relative humidity (RH), reverting back to the original local order as the RH
decreased. Phase-separation in the miscible dispersions as a result of ambient RH conditions did
not appear to take place. Finally, formation of pores (voids) was observed through small-angle
X-ray powder diffraction during crystallization of one model drug (felodipine). ß 2010 Wiley-Liss,
Inc. and the American Pharmacists Association J Pharm Sci 99:4005 4012, 2010
Keywords: amorphous; solid dispersion; crystallization; physical stability; X-ray powder
diffractometry; polymeric drug delivery systems; stabilization
INTRODUCTION and various hydroxyproylmethyl cellulose (HPMC)
and polyacrylic acid (PAA) derivatives, to form
Amorphous solid dispersions of active pharmaceuti- what are described as   miscible  solid dispersions,
cal ingredients (APIs) and pharmaceutically accep- as opposed to phase-separated mixtures of two or
table polymers have been a topic of interest in recent more components.7   Miscibility  , in this sense, does
years for the solid-state community due to their not mean an API polymer mixture in which the API
potential in improving oral bioavailability, especially exhibits equilibrium solubility in the polymer. Rat-
for drugs exhibiting poor aqueous solubility.1 3 Where her, the definition describes a system consisting of a
dissolution rate can be identified as the rate-limiting single supersaturated metastable phase of API and
step in oral absorption, one can expect poor aqueous polymer, where the components are able to mutually
solubility to have a negative impact on oral bioavail- influence the solid structure of the other phase. Pro-
ability.4 It is generally true that amorphous forms perties of such systems are typically significantly
have increased free energy, solubility (in any solvent), altered as compared to starting materials, affecting
and dissolution rate (in any solvent), but also molecular mobility and intrinsic properties, including
increased chemical and thermodynamic activity, as the tendency to recrystallize or to undergo chemical
compared to crystalline polymorphs.5 These increa- instability over practical storage times. Important
sed properties typically lead to (often orders of thermodynamic criteria for the formation of meta-
magnitude) greater oral absorption and bioavailabil- stable amorphous molecular dispersions have been
ity of amorphous materials, but also significantly previously studied8 and include a sufficiently positive
greater chemical and physical instability.6 combinatorial entropy and the ability to form inter-
One approach typically used to overcome instability molecular interactions, for example, through hydro-
problems with amorphous APIs is to combine them gen bonding or dipole dipole interactions. Clearly,
with pharmaceutically acceptable polymers, such such API polymer interactions need to be sufficiently
as polyvinylpyrrolidone (PVP), polyvinylpyrrolidone favored to offset interactions between the individual
vinyl acetate (PVP/VA), polyethylene glycol (PEG), components (i.e., API API and polymer polymer) in
the mixture.
Miscible dispersions are usually characterized by
Correspondence to: Igor Ivanisevic (Telephone: [phone];
a single glass transition temperature intermediate
Fax: [phone]; E-mail: [e-mail])
to those of the polymer and API, as opposed to two
Journal of Pharmaceutical Sciences, Vol. 99, 4005 4012 (2010)
ß 2010 Wiley-Liss, Inc. and the American Pharmacists Association observed glass transition temperatures corresponding
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010 4005
4006 IVANISEVIC
to the API and polymer phases, in a phase separated some of) the systems in question. We are not aware of
system. It has been noted in literature that, due to other published long-term ambient stability studies
limitations of differential scanning calorimetry (DSC), on miscible dispersions.
it is possible to observe a single glass transition tem-
perature in a phase-separated system and vice
EXPERIMENTAL
versa.9 11 Furthermore, heat (and humidity) have
been shown to have an impact on the miscibility of a
Materials
system,12,13 suggesting the possibility of sample alter-
ation during the course of a DSC measurement. Dichloromethane (ChromAR grade) and chloroform
Therefore, nondestructive, structure-based techni- (AR grade) were obtained from Mallinckrodt Baker,
ques built around the use of X-ray powder diffraction Inc. (Paris, KY), while ethanol (200 proof) was
(XRPD) and computational methods have been rec- obtained from Pharmco-AAPER (Shelbyville, KY).
ently developed to provide a complementary assess- Felodipine was a generous gift from AstraZeneca
ment of miscibility in an amorphous dispersion.11,14,15 (Södartälje, Sweden). Polyvinyl pyrrolidone (PVP)
Such techniques detect the changes in the local order K29-32 and PAA (Average Mv 450,000) were purc-
present in amorphous dispersions as compared to hased from Sigma Aldrich Co. (St. Louis, MO).
local order in the amorphous API and excipient comp- Ketoconazole and nifedipine were obtained from
onents hypothesized to occur as a result of inter- Hawkins, Inc (Minneapolis, MN). Prior to use, PVP
action between an API and excipient. and PAA were dried in a desiccator over powdered
Where miscibility can be achieved, physical sta- phosphorus pentoxide for at least 1 week.
bilization is thought to take place by reduction of
molecular mobility of the API molecules and/or by Bulk Sample Preparation
inhibition of nucleation and crystal growth through
Binary mixtures of the model drug and polymer
preferential API polymer interactions.16,17 Numer-
were prepared at different weight ratios, and then
ous examples of miscible dispersions have been rep-
dissolved in a common solvent. For FEV PVP and
orted in literature,18 21 primarily based on DSC and
nifedipine PVP systems, the solvent was a 1:1 (w/w)
spectroscopic characterization. The physical stability
mixture of dichloromethane and ethanol. For KET
of such systems has been studied, typically over short-
PVP systems, chloroform was used, while for FEL
to-intermediate timeframes under predefined tempera-
PAA, the solvent was pure ethanol. For IMC PVP,
ture and relative humidity (RH) conditions.20,22,23 To
the solvent was dichloromethane. All mixtures
the best of our knowledge, no long-term studies of
were visually inspected to confirm that the drug
systems where miscibility has been inferred through
and polymer were completely dissolved, and the
structure-based techniques have been published to
systems formed uniform one-phase solutions. The
date. Such data would enhance the confidence in
solvent was then removed using a rotary evaporator
structural techniques as a predictor of physical
apparatus (Brinkman Instruments, Westbury, NY),
stability in amorphous dispersions, and encourage
and the samples were subsequently placed under
their use for routine dispersion screening.
vacuum for at least 12 h to remove any residual
We present a study of 12 model dispersion systems
solvents. Additional observations related to sample
over 9 22 months. The miscibility of these systems
preparation can be found in previously published
was previously assessed using structure-based (as well
reports.11,15
as spectroscopic and thermal) techniques,11,14,15 and
we believe the results directly correlate to the phy-
Storage
sical stability data presented in this report. The
systems include ketoconazole PVP (KET PVP; at Samples were stored in a light-protected chamber,
30% and 70% API), felodipine PVP (FEV PVP; at sandwiched between two sheets of thin (3 mm)
30% and 70% API), nifedipine PVP (at 30%, 40%, Etnom1 (Chemplex, Palm City, FL) film. Each
60%, and 70% API), felodipine PAA (FEL PAA; at sandwich was packed in an XRPD holder and the sam-
40% and 60% API) and indomethacin PVP (IMC ples remained packed in this fashion for the duration
PVP; at 40% and 50% API). The physical stability of of the study and during all XRPD measurements.
individual amorphous APIs (not stabilized by exci- Etnom1 is nearly XRD-transparent, and the (mini-
pient) under the same conditions was also studied, for mal) film contribution to X-ray diffraction was asse-
comparison. Ambient (but monitored) temperature ssed through the use of blanks. Storage temperature
and humidity conditions were chosen for the study. and humidity were monitored hourly and varied
Other groups have reported effects of (higher) temp- seasonally. The temperature ranged from 22.7 to
erature and humidity stress on some of these 26.58C (average about 248C), while RH ranged from
dispersion systems;12,13 temperature and humidity 2% to 62%, averaging 35% RH over the length of the
are known to induce phase separation in (at least study. Table 1 provides detailed temperature and RH
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010 DOI 10.1002/jps
STUDIES OF MISCIBLE AMORPHOUS SOLID DISPERSIONS 4007
Table 1. Sample Storage Conditions
Temperature (8C) Relative Humidity (%)
Duration
Sample Description (Days) Min Max Avg s Min Max Avg s
FEL #1 Amorphous API 11a 23.2 24.9 24.0 0.4 41 54 48 3
FEL #2 Amorphous API 3a 24.2 25.0 24.6 0.2 42 54 47 3
FEL #3 Amorphous API 305a 23.0 26.5 24.2 0.5 2 62 33 16
FEL 30 PVP 70 Miscible dispersion 441 22.9 26.5 24.3 0.5 2 62 36 14
FEL 70 PVP 30 Miscible dispersion 673 22.7 26.5 24.1 0.5 2 62 35 13
FEL 40 PAA 60 Phase-separated dispersion 331a 22.9 26.5 24.2 0.5 2 59 31 13
FEL 60 PAA 40 Phase-separated dispersion 35a 22.9 25.4 24.2 0.3 30 59 48 5
NIF #1 Amorphous API 14a 23.2 25.1 24.1 0.5 41 54 49 3
NIF #2 Amorphous API 2a 24.2 25.0 24.6 0.2 42 56 49 4
NIF #3 Amorphous API 2a 24.4 25.2 24.7 0.2 36 52 42 4
NIF 30 PVP 70 Miscible dispersion 267 23.0 26.5 24.2 0.4 2 62 37 17
NIF 40 PVP 60 Miscible dispersion 404 22.9 26.5 24.2 0.5 2 62 35 14
NIF 60 PVP 40 Miscible dispersion 396 22.9 26.5 24.2 0.5 2 62 35 14
NIF 70 PVP 30 Miscible dispersion 662b 22.7 26.5 24.1 0.5 2 62 35 13
KET Amorphous API 416a 22.9 26.5 24.2 0.5 2 62 36 14
KET 30 PVP 70 Miscible dispersion 419 22.9 26.5 24.2 0.5 2 62 36 14
KET 70 PVP 30 Phase-separated dispersion 642a 22.7 26.5 24.1 0.5 2 62 35 13
IMC 40 PVP 60 Miscible dispersion 460 22.9 26.5 24.3 0.5 2 62 37 14
IMC 50 PVP 50 Miscible dispersion 687 22.7 26.5 24.1 0.5 2 62 35 13
a
Sample crystallized during study.
b
Sample initially contained minor amounts of crystalline material. Over time, a small portion of the sample (estimated <10% by XRPD) crystallized.
conditions for each sample over the duration of each approximately 60 min, for each pattern and time-
study. point.
Computational Analysis
DSC
The computational tools used to analyze these dis-
Modulated DSC data were obtained on a TA Instru-
persions were described in earlier publications11,14
ments (New Castle, DE) Q2000 differential scanning
and the results of their application to these model
calorimeter equipped with a refrigerated cooling sys-
systems were also reported previously.11,14,15 The
tem (RCS). Temperature calibration was performed
tools are implemented in a proprietary software
using NIST traceable indium metal. The sample was
package, PatternMatch v3.0.0.24
placed into an aluminum DSC pan, and the weight
XRPD was accurately recorded. The pan was covered with a
lid, and the lid was crimped. A weighed, crimped
XRPD patterns were collected using a PANalytical
aluminum pan was placed on the reference side of the
(Almelo, The Netherlands) X Pert Pro diffractometer.
cell. Data were obtained using a modulation ampli-
An incident beam of Cu Ka radiation was produced
tude of 0.88C and a 60 s period with an underlying
using an Optix long, fine-focus source. An elliptically
heating rate of 18C/min from 25 to 1508C. The
graded multilayer mirror was used to focus the Cu Ka
reported glass transition temperatures are obtained
X-rays of the source through the specimen and onto
from the inflection of the step change in the reversing
the detector. Data were collected and analyzed using
heat flow versus temperature curve.
X Pert Pro Data Collector software (v. 2.2b). Prior to
the analysis, a silicon specimen (NIST SRM 640c)
was analyzed to verify the Si 111 peak position. The RESULTS AND DISCUSSION
specimen was sandwiched between 3-mm thick films,
analyzed in transmission geometry, and rotated All dispersion samples were monitored monthly for
to optimize orientation statistics. A beam-stop and changes in the local order using XRPD. The pure
a helium atmosphere were used to minimize the amorphous (reference) drug samples were also analy-
background generated by air scattering. Soller slits zed by XRPD under similar conditions, to provide a
were used for the incident and diffracted beams to baseline for comparison. The study conditions for
minimize axial divergence. Diffraction patterns were each sample, that is, time, temperature, and humid-
collected using a scanning position-sensitive detector ity are listed in Table 1.
(X Celerator) located 240 mm from the specimen. The nine dispersions initially characterized as mis-
The scan range was 1 988 2u and collection time cible using structure-based techniques all remained
DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
4008 IVANISEVIC
X-ray amorphous for the duration of the studies (up to
22 months). Slight crystallization (<10% by bulk)
appeared evident over time in the case of nifedipine
PVP (at 70% drug loading); XRPD analysis detected
no signs of crystallization in any of the other miscible
dispersions studied. This was in contrast to the beha-
vior of the pure amorphous drugs used in these
studies; felodipine and nifedipine crystallized over a
few days while ketoconazole partially crystallized in a
matter of weeks, under the same conditions. The
three dispersion systems initially characterized as
phase-separated (KET PVP at 70% API, and FEL
PAA at 40 and 60% API loading) all crystallized over a
period of 1 2 months. Detailed analysis of the results
for each system is presented below.
Felodipine PVP (FEL PVP)
Figure 1. XRPD analysis of the felodipine (70%) polyvi-
nylpyrrolidone (30%) dispersion, over time.
Two samples prepared at 30% and 70% felodipine
loading, respectively were included in the study.
XRPD and computational analysis indicated changes
in the local order of amorphous felodipine and PVP order of the dispersion took place during 50% RH
upon initial dispersion preparation.15 The results of summer months. As the RH dropped to about 20 30%
the computational analysis for this and other samples in the fall and winter, additional changes in the local
can be found in the cited references and it should be order were observed in the XRPD patterns. The halos
noted that the samples used for characterization shifted back towards their original positions, centered
were the same samples used for this stability study. at 11.3 and 218 2u. Thereafter, no further changes in
Therefore, results of the computational analysis the local order were observed for this sample for the
are omitted in this manuscript. The computational remainder of the study. In comparison, an XRPD
result was consistent with spectroscopy observations pattern of pure amorphous felodipine contains halo-
of specific interactions between felodipine and PVP, centered at 11.6 and 238 2u, while XRPD data of PVP
persisting over a wide range of drug concentra- K29-32 contain halos at 11.6 and 20.08 2u (Fig. 3).
tions.13,15 Therefore, these two dispersions were The changes in the local order observed in the 30%
thought to be miscible following initial preparation. felodipine dispersion over time could potentially be
Published reports on FEV PVP dispersions suggest explained by the phase separation effect of humidity
this system undergoes phase separation upon humid- reported previously.12,13 However, that would not
ity stress 75% RH.12,13 The RH conditions for the explain the lack of such changes in the 70% felodipine
duration of our study never exceeded 62% (Tab. 1). dispersion stored under the same conditions and
For the 70% felodipine dispersion, no changes in the timeframe. Furthermore, the humidity in our study
local order were detected by XRPD for the duration of
the study, about 22 months (Tab. 1, Fig. 1). The Tg of
this material was approximately 668C upon initial
preparation. Despite the relatively low Tg, this mis-
cible dispersion did not crystallize to within the
detection limit of the XRPD technique used to analyze
the sample (estimated around 1% by mass). In
addition, no evidence of amorphous phase separation
could be observed in the X-ray data; that is, there
were no shifts in the position of the X-ray amorphous
halos over time.
The 30% felodipine dispersion (Tg ź 958C) likewise
did not crystallize over the 14 month study. However,
some variability in the XRPD data was observed,
suggesting structural changes in the sample over
time (Fig. 2). The low angle halo in the XRPD patterns
collected on this dispersion initially shifted over time
from 11.6 to 10.78 2u, while the high-angle halo shifted
Figure 2. XRPD analysis of the felodipine (30%) polyvi-
from 20.9 to 21.88 2u. These initial changes in the local nylpyrrolidone (70%) dispersion, over time.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010 DOI 10.1002/jps
STUDIES OF MISCIBLE AMORPHOUS SOLID DISPERSIONS 4009
Figure 3. XRPD analysis of amorphous PVP (top) and Figure 4. Crystallization of amorphous felodipine over
felodipine (bottom). time.
never reached 75% RH, reported to be the lower limit While analyzing the amorphous felodipine samples,
for moisture-induced phase separation in this system. an apparent rise in the small-angle X-ray scattering
Finally, the 30% dispersion did not crystallize, as (SAXS) intensity was observed during crystallization.
would be expected from a phase-separated dispersion The XRPD instrument was reconfigured for SAXS
of felodipine and PVP over such a long-time period. measurements, and concurrent wide-angle X-ray
Another possible explanation for the changes obser- scattering (WAXS) and SAXS analyses were collected
ved is the natural aging of what is a thermodynami- on one crystallizing sample. The WAXS data indi-
cally unstable system. Even at mild environmental cated crystallization of felodipine Form I (Fig. 4). The
conditions, one would expect some irreversible chan- SAXS data contained an initially gradually intensify-
ges to take place as a result of aging. However, if such ing SAXS peak centered at approximately 18 2u
aging was the sole reason for the differences observed (Fig. 5). The intensity of this peak leveled off after
in the X-ray data, one would not expect to see the halo about 24 h and then persisted in the crystallized mat-
positions retrace back towards their original values erial for the length of the observation (15 months).
following the initial shift, apparently as a function of This result was reproduced on multiple samples of
seasonal RH. Also, aging should eventually result felodipine. The same peak was not observed in X-ray
in crystallization which was likewise not observed. results collected from as-received crystalline felodi-
While we cannot rule out aging being part of the pine obtained from commercial sources.
reason for the observed phenomenon, it does not
appear to be the sole reason.
Therefore, it is our hypothesis that the observed
changes in the local order were at least in part the
result of moisture absorption by PVP,12 which was
likely present in excess of what was required to form a
miscible dispersion with the felodipine present. This
hypothesis is supported by results from a separate
study of structural changes in PVP as a function of
RH,25 where similar halo shifts were observed.
Three samples of pure amorphous felodipine were
prepared using the same methods and stored under
similar conditions (Tab. 1) as the dispersions and
monitored by XRPD in order to provide a baseline for
comparison of physical stability. All three samples
fully crystallized inside their Etnom1 sandwiches
within 1 week (Fig. 4). Therefore, physical stability of
amorphous felodipine under ambient conditions was
increased from less than 1 week to at least 22 months
Figure 5. SAXS of crystallizing felodipine over time (data
by dispersing the material with 30% or 70% PVP. smoothed).
DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
4010 IVANISEVIC
Additional analysis of crystalline felodipine Form I, crystallization observed in the 70% nifedipine sample
with and without the SAXS peak, was undertaken was related to a small portion of the sample that was
using atomic force microscopy, isothermal calorime- initially not stabilized by excipient. Once this portion
try, dynamic vapor sorption, DSC, and scanning elect- of the sample crystallized, no further crystallization
ron microscopy. None of these techniques could detect appeared to take place in the bulk of the sample. This
any appreciable difference in physical properties of result suggests that miscible dispersions can exhibit
these two crystalline materials. It is hypothesized resistance towards crystallization even in the pre-
that felodipine crystallized from amorphous material sence of seeds of crystalline API material.
likely contained voids (pores), similar to previously The seasonal local order changes previously dis-
published reports for inorganic materials.26,27 The cussed for the 30% felodipine dispersion were also
diffraction from the electron density differences observed in the 30% and, to a lesser extent, 40%
in pores versus solid surfaces likely resulted in the nifedipine dispersions. Once again, the changes in
SAXS peak observed in XRPD results. From the XRPD patterns appeared to correlate to RH condi-
position of the SAXS peak, the size of the voids was tions, with halo shifts in high-RH months that
calculated to be approximately 100 Å, but this value reverted towards their original position in low-RH
should be taken as a rough estimate due to uncert- months. Three samples of amorphous nifedipine were
ainties associated with the SAXS measurements. monitored under similar storage conditions to provide
reference crystallization data. All three samples fully
Nifedipine PVP
crystallized within approximately 2 days.
Four samples, prepared at 30%, 40%, 60%, and 70%
Ketoconazole PVP (KET PVP)
API loading, were included in the study. As with
felodipine dispersions, all four of these systems were Two samples, prepared at 30% and 70% API loading,
initially characterized as miscible amorphous disper- were included in the study. Evidence of changes in
sions using structure-based, thermal, and spectro- the local order of the 30% API sample were observed
scopic techniques.15 The samples were placed on by XRPD upon initial preparation, suggesting mis-
stability and monitored by XRPD monthly (Tab. 1). cibility. At the higher loading, no clear evidence of
Some residual crystallinity was observed by XRPD structural changes could be observed and phase-
in the 70% dispersion sample immediately upon pre- separation was suspected.15 Spectroscopic analysis
paration (Fig. 6). Minor crystallization was observed for these systems was inconclusive, due to lack of
for this sample over the first 6 months of the study; hydrogen bonding potential; only a single Tg was
thereafter, the sample did not appear to crystallize observed at both API loadings.15
further. At the final timepoint (22 months), this The 30% API sample did not crystallize over the
sample still contained less than 10% crystalline length of the study (about 1 year), although changes
nifedipine as estimated from XRPD results (Fig. 6). in the local order apparently correlated to seasonal
No crystallization was observed for the other three variability in RH conditions were observed for this
nifedipine dispersion samples, including the 60% API sample, similar to the results for the 30% felodipine
sample. Therefore, it is our hypothesis that the minor and nifedipine dispersions. Crystallization of the 70%
API sample began after about 2 months of storage and
very gradually progressed for the remainder of the
study (18 months, Fig. 7). Therefore, the miscible
ketoconazole dispersion demonstrated physical sta-
bility that was at least six times longer than the
phase-separated ketoconazole dispersion under the
same conditions. In addition, pure amorphous keto-
conazole was prepared and stored for over a year
under the same conditions. The material began to
crystallize after 8 days, though it remained partially
X-ray amorphous at the end of the study (416 days).
Indomethacin PVP (IMC PVP)
Two samples, prepared at 40% and 50% API loading,
were included in the study. The miscibility of this
system was previously reported in literature based on
spectroscopic,19 thermal, and structure-based tech-
niques.11,14 No crystallization or clear changes in the
Figure 6. XRPD analysis of the nifedipine (70%) poly- local order could be detected for these samples over
vinylpyrrolidone (30%) dispersion, over time. the duration of the study (Fig. 8). However, some
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010 DOI 10.1002/jps
STUDIES OF MISCIBLE AMORPHOUS SOLID DISPERSIONS 4011
Figure 9. XRPD analysis of the felodipine (40%) poly-
Figure 7. XRPD analysis of the ketoconazole (70%) poly-
acrylic acid (60%) dispersion, over time.
vinylpyrrolidone (30%) dispersion, over time.
variation in the intensity of the first amorphous halo- The 40% sample was kept on stability for approxi-
centered at approximately 118 2u was observed, mately 1 year and crystallized further over time
which could correspond to changes in the local order (Fig. 9).
over time. For comparison, pure amorphous indo-
methacin prepared under the same conditions was
partially crystalline immediately following prepara- CONCLUSION
tion.
The study outlined in this report demonstrates that
miscible amorphous solid dispersions exhibit very
Felodipine PAA (FEL PAA)
significant physical stability with respect to crystal-
Two samples, prepared at 40% and 60% API loading,
lization under ambient conditions. Of the nine mis-
were included in the study. XRPD and spectroscopic
cible systems studied FEL PVP at 30 and 70% API,
analysis of as-prepared samples indicated the sys- NIF PVP at 30%, 40%, 60%, 70% API, KET PVP at
tems were phase-separated, while thermal analysis 30% API, and IMC PVP at 40% and 50% API only
detected two or one Tgs, respectively.15 Storage of the 70% NIF 30% PVP system showed any detectable
these samples resulted in crystallization of felodipine,
(by XRPD) signs of crystallization over the length of
observed at the first timepoint (after about 1 month). the respective studies (9 22 months). For that sys-
tem, bulk crystallization over 22 months amounted to
less than 10%.
In contrast, all three phase-separated systems
KET PVP at 70% API, FEL PAA at 40% and 60%
API began to noticeably crystallize within 1 2
months. Furthermore, the pure amorphous drugs
used in this study all crystallized within a few days.
Therefore, miscible dispersions have been shown to
increase the shelf-life of otherwise fast-crystallizing
amorphous drugs by orders of magnitude under
ambient conditions. These results suggest dispersion
design with the focus on achieving miscibility can
result in formulations suitable not only for early
phase studies (toxicity, efficacy) but potentially for
final formulation. The results further suggest that
structure-based characterization, used to detect   mis-
cibility  in a system through changes in the local
order present in amorphous materials, may poten-
Figure 8. XRPD analysis of the indomethacin (50%) tially be used as a predictor of long-term physical
polyvinylpyrrolidone (50%) dispersion, over time. stability of amorphous dispersion systems. Whenever
DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010
4012 IVANISEVIC
possible, complementary techniques (e.g., spectro- 10. Krause S, Iskander M. 1977. Phase separation in styrene-a-
methyl styrene block copolymers. In Polymer Science and
scopic or thermal) should also be used to characterize
Technology. D. Klempner, & K.C. Frisch, Eds. New York:
such systems to gain increased confidence in the
Plenum Press, Vol. 10: pp. 231 243.
physical stability predictions.
11. Newman A, Engers D, Bates S, Ivanisevic I, Kelly R, Zografi G.
Finally, the report presents a pair of observations
2008. Characterization of API:polymer mixtures using X-ray
on the structure of amorphous materials. The first powder diffraction. J Pharm Sci 97:4840.
12. Rumondor ACF, Marsac PJ, Stanford L, Taylor LS. 2009. Phase
was related to changes in the local order in amorp-
behavior of poly(vinylpyrrolidone) containing amorphous solid
hous dispersions over time, apparently as a result of
dispersions in the presence of moisture. Mol Pharm 6:1492
humidity effects on PVP. While these changes do
1505.
not appear to be a consequence of humidity-induced
13. Marsac PJ, Rumondor ACF, Nivens DE, Kestur US, Stanciu L,
phase separation and do not result in API crystal- Taylor LS. 2009. Effect of temperature and moisture on the
miscibility of amorphous dispersions of felodipine and poly(-
lization, they may have an effect on the in vivo or in
vinylpyrrolidone). J Pharm Sci 99:169 185.
vitro performance of such dispersions. The second
14. Ivanisevic I, Bates S, Chen P. 2009. A novel method for the
observation was related to apparent formation of
assessment of miscibility in amorphous dispersions. J Pharm
voids (pores) in felodipine as the material crystallizes
Sci 98:3373 3386.
from the amorphous state. The resulting crystalline 15. Rumondor ACF, Ivanisevic I, Bates S, Alonzo DE, Taylor LS.
2009. Evaluation of drug polymer miscibility in amorphous
material still contains the voids, although their
solid dispersion systems. Pharm Res 26:2523 2534.
presence could not be correlated to a change in any
16. Marsac PJ, Li T, Taylor LS. 2009. Estimation of drug polymer
physical property, as compared to crystalline felodi-
miscibility and solubility in amorphous solid dispersions using
pine that did not contain such voids.
experimentally determined interaction parameters. Pharm Res
26:139 151.
17. Van den Mooter G, Wuyts M, Blaton N, Busson R, Grobet P,
ACKNOWLEDGMENTS
Augustijns P, Kinget R. 2001. Physical stabilisation of amor-
phous ketoconazole in solid dispersions with polyvinylpyrroli-
done K25. Eur J Pharm Sci 12:261 269.
The author wishes to thank Mark Andres and Alfred
18. Chiou WL, Riegelman S. 1969. Preparation and dissolution
Rumondor for help with sample preparation and Simon
characteristics of several fast-release solid dispersions of
Bates and Lynne Taylor for helpful discussions.
griseofulvin. J Pharm Sci 58:1505 1510.
19. Taylor LS, Zografi G. 1997. Spectroscopic characterization of
interactions between PVP and indomethacin in amorphous
REFERENCES
molecular dispersions. Pharm Res 14:1691 1698.
20. Aso Y, Yoshioka S, Kojima S. 2003. Molecular mobility-based
1. Hancock BC. 2002. Disordered drug delivery: Destiny, estimation of the crystallization rates of amorphous nifedipine
dynamics and the deborah number. J Pharm Pharmacol 54: and phenobarbital in poly(vinylpyrrolidone) solid dispersions.
737 746. J Pharm Sci 93:384 391.
2. Leuner C, Dressman J. 2000. Improving drug solubility for oral 21. Marsac PJ, Konno H, Taylor LS. 2006. A comparison of the
delivery using solid dispersions. Eur J Pharm Biopharm 50: physical stability of amorphous felodipine and nifedipine sys-
47 60. tems. Pharm Res 23:2306 2316.
3. Craig DQM. 2002. The mechanism of drug release from solid 22. Matsumoto T, Zografi G. 1999. Physical properties of solid
dispersions in water soluble polymers. Int J Pharm 231:131 molecular dispersions of indomethacin with poly(vinylpyrroli-
144. done) and poly(vinylpyrrolidone-co-vinyl-acetate) in relation to
4. Yalkowsky SH. 1999. Solubility and solubilization in aqueous indomethacin crystallization. Pharm Res 16:1722 1728.
media. 1st edition. Oxford: Oxford University Press. 23. Yang J, Grey K, Doney J. 2010. An improved kinetics approach
5. Grant DJW, Higuchi T. 1990. Solubility behavior of organic to describe the physical stability of amorphous solid disper-
compounds. 1st edition. New York, NY: John Wiley & Sons. sions. Int J Pharm 384:24 31.
6. Hancock BC, Zografi G. 1997. Characteristics and significance 24. Ivanisevic I, Bugay DE, Bates S. 2005. On pattern matching
of the amorphous state in pharmaceutical systems. J Pharm of X-ray powder diffraction data. J Phys Chem B 109:7781
Sci 86. 7787.
7. Serajuddin ATM. 1999. Solid dispersion of poorly water soluble 25. Teng J, Bates S, Engers DA, Leach K, Schields P, Yang Y. 2010.
drugs: Early promises, subsequent problems, and recent break- Effect of water-vapor sorption on local structure of polyvinyl-
throughs. J Pharm Sci 88:1058 1066. pyrrolidone. J Pharm Sci (accepted).
8. Shamblin SL, Taylor LS, Zografi G. 1998. Mixing behavior of co- 26. Jansen M, Schön JC, van Wüllen L. 2006. The route to the
lyophilized binary amorphous systems. J Pharm Sci 87:694 structure determination of amorphous solids: A case study
701. of the ceramic Si3B3N7. Angew Chem, Int Ed Engl 45:4244
9. Lodge TP, Wood ER, Haley JC. 2006. Two glass transitions do 4263.
not necessarily indicate immiscibiity: The case of PEO/PMMA. 27. Himmel B, Gerber T, Heyer W, Blau W. 1987. X-ray diffraction
J Polym Scir B, Polym Phys 44:756 763. analysis of SiO2 structure. J Mater Sci 22:1374 1378.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 9, SEPTEMBER 2010 DOI 10.1002/jps