The influence of water on the stability of lyophilized formulations with inositol and mannitol as excipients

Article abstract of DOI:10.1248/cpb.57.459

The stability of a moisture-sensitive drug in lyophilized products was investigated under conditions with varying water content and temperature using two model formulations: a formulation containing inositol (IF) as the excipient and a formulation contain

Full text of DOI:10.1248/cpb.57.459

May 2009
Chem. Pharm. Bull. 57(5) 459—463 (2009)
459
The Influence of Water on the Stability of Lyophilized Formulations with
Inositol and Mannitol as Excipients
,
a
a
b
Akira TERAKITA,* Hirokazu MATSUNAGA, and Tetsuro HANDA
a
Analytical Development Laboratories, Takeda Pharmaceutical Co., Ltd.; 2–17–85 Jusohonmachi, Yodogawa-ku, Osaka
32–8686, Japan: and Graduate School of Pharmaceutical Sciences, Kyoto University; 46–29 Yoshida-Shimo-Adachi-
b
5
cho, Sakyo-ku, Kyoto 606–8501, Japan. Received September 13, 2008; accepted February 2, 2009
The stability of a moisture-sensitive drug in lyophilized products was investigated under conditions with
varying water content and temperature using two model formulations: a formulation containing inositol (IF) as
the excipient and a formulation containing mannitol (MF) as the excipient. IF showed better chemical stability (a
lower hydrolysis rate) than MF when both formulations contained 2% water. However, in the case of formula-
tions with 8% water, MF showed similar or better stability than IF. From the results of hygroscopicity and phase
transition experiments for both formulations, it was assumed that this stability profile was exhibited because 1)
more water was taken up into the amorphous inositol in IF than into the crystalline Form-III mannitol in MF at
a low water content, so that drug hydrolysis in IF was suppressed compared with MF and 2) when the water con-
tent increased, the amorphous inositol crystallized to anhydrate in IF causing expulsion of absorbed water from
the excipient, meaning that IF lost its superior chemical stability due to the highly mobile water generated by the
2
crystallization. This assumption was supported by the results of the H-NMR measurement, which estimated
water mobility from the signal shape and the spin–lattice relaxation time (T ) of deuterium oxide.
1
2
Key words lyophilization; stability; phase transition; water mobility; hygroscopicity; H-NMR
Lyophilization is widely used in manufacturing pharma- 2-(2,4-difluorophenyl)-2-hydroxy-3-[2-oxo-3-[4-(1H-1-tetra-
ceutical dosage forms, especially in the case of drug sub- zolyl)phenyl]-1-imidazolidinyl]butyl]-1H-1,2,4-triazolium
6)
stances that are susceptible to chemical degradation in liquid chloride (TAK-457), a proprietary active ingredient, and
formulations. Although both chemical and physical stability commonly used bulking agents, was measured at 25 °C and
problems in lyophilized formulations have been known for 40 °C. TAK-457 was developed as an injectable prodrug of
many years, in some cases less attention has been paid to the 1-[(1R,2R)-2-(2,4-difluorophenyl)-2-hydroxy-1-methyl-3-
solid-state properties of active pharmaceutical ingredients (1H-1,2,4-triazol-1-yl)propyl]-3-[4-(1H-1-tetrazolyl)phenyl]-
6)
and excipients in such formulations than to those in tablet 2-imidazolidinone (TAK-456) with improved water solubil-
6)
and capsule formulations. For example, conversion between ity against fungal infections (Fig. 1). This study is unique
crystalline and amorphous forms can greatly affect the stabil- and helps us reveal the influence of residual water, tempera-
1—3)
ity and solubility of drugs in lyophilized formulations.
ture, and excipients on stability, because TAK-457 degrades
Residual moisture may have a significant influence on the to TAK-456 predominantly via hydrolysis in its solid state
stability and quality of formulations because water can act and the amounts of other degradation products are negligible.
not only as a chemical reactant but also as a plasticizer to fa- Two model formulations of TAK-457 containing inositol or
2)
cilitate the crystallization of amorphous materials. Of more mannitol as the excipient were used. Hygroscopicity, phase
significance for drug stability, as some researchers have transition, and water mobility in these formulations were es-
demonstrated, is the location of water molecules in the solid timated in order to investigate the differences in the chemical
2
matrix and the nature of water–solid interactions rather than stability of the drug. Water mobility was determined by H-
2—4)
the amount of water present.
When water acts as a plasti- NMR.
cizer, it increases the mobility of the system, leading to en-
hanced reactivity. NMR is a powerful method for determin-
ing mobility in solids. Aso et al. estimated water mobility in
Experimental
Materials TAK-457 was manufactured at Takeda Pharmaceutical Co.,
Ltd. (Osaka, Japan). D(ꢀ)-Mannitol of JIS special grade was purchased from
Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Inositol was purchased
from Shiratori Pharmaceutical Co., Ltd. (Chiba, Japan).
2
solid samples by solid-state H-NMR and investigated the in-
4,5)
fluence of mobility on drug stability.
Sample Preparation Lyophilized formulations were obtained by plac-
In this study the stability of freeze-dried products contain- ing vials containing 30.6 mg of TAK-457 with 12.5 mg of inositol or manni-
ing the water-soluble prodrug, 4-acetoxymethyl-1-[(2R,3R)- tol in 2.5 ml of aqueous solution onto the shelf of a freeze dryer. This was
Fig. 1. Chemical Structures of TAK-457 and TAK-456
∗
To whom correspondence should be addressed. e-mail: Terakita_Akira@takeda.co.jp
© 2009 Pharmaceutical Society of Japan

4
60
Vol. 57, No. 5
followed by freezing at ꢀ50 °C for 4 h and then primary drying at a shelf 25 °C and 40 °C. The drug contents in the initial and stored
temperature of ꢀ10 °C and a chamber pressure of 60 mTorr for 22 h. Sec-
ondary drying was then carried out at a shelf temperature of 45 °C and a
chamber pressure of 60 mTorr for 6 h. Lyophilized samples with 2% and 8%
water were prepared for the stability study by storing the formulations under
samples were measured by HPLC. The content decreased
during storage predominantly via hydrolysis. The main
degradation product of TAK-457 was TAK-456, as shown in
a constant humidity of about 53% relative humidity (RH) or 75% RH at am- the representative chromatograms (Fig. 2). Even when the
bient temperature (about 25 °C) and monitoring their weight increase.
HPLC Analysis A liquid chromatograph was equipped with a 275-nm
detector and a 4.6ꢁ150 mm YMC-Pack Pro C18 column (YMC Co., Ltd.,
Kyoto, Japan) and was operated at a constant temperature of 25 °C. The mo-
bile phase was a mixture of 0.05 mol/l sodium dihydrogenphosphate solution
drug content decreased to about 20% after 14-d storage of
the samples with 8% water at 25 °C, the amounts of other
degradation products were negligible (below 0.5%).
In the comparison between IF and MF samples containing
and acetonitrile (13 : 7). The flow rate was 1.0 ml/min. The sample solution 2% water, as shown in Fig. 3A, IF showed better stability at
was prepared by dissolving the contents of a vial in the mobile phase. The both 25 °C and 40 °C. However, in the case of samples with
concentration of TAK-457 in the vial was determined by the calibration
8
3
% water, MF showed similar or slightly better stability (Fig.
B). Interestingly, not only was the hydrolysis reaction of the
method using a standard solution of TAK-457.
Karl Fischer Titrimetry (KFT) The initial water content was deter-
mined using an Aquacounter AQ-7 (Hiranuma Sangyo Co., Ltd., Ibaraki,
drug significantly affected by the excipients, but the stability
Japan). The contents of a vial were dissolved in 2 ml of methanol. Precise 1- between the formulations also changed due to the water con-
ml portions of this methanol solution were transferred into a titration cell tent, even though both formulations were manufactured
with Aqualyte RS as the anolyte and Aqualyte CN as the catholyte (Hi-
ranuma Sangyo Co., Ltd., Ibaraki, Japan).
Sorption and Desorption Isotherm Isothermal sorption and desorp-
tion profiles were measured using the SGA-100 Symmetric Vapor Sorption
under the same lyophilization conditions using solutions con-
taining the same amounts of the drug and the excipients (al-
most the same molar concentrations of excipient). Some re-
Analyzer, a humidity-controlled microbalance system (VTI Corporation, ports have indicated that drug degradation rates could be in-
Hialeah, FL, U.S.A.). Sorption and desorption data were collected over a
range of 5 to 95% RH at 5% RH intervals under a nitrogen purge. Samples
were dried at 25 °C for up to 2 h prior to analysis with the equilibrium crite-
ria defined as a weight change ꢂ0.01% in the 2 min. The weight change was
monitored at each RH for a maximum time of 4 h allowed for the samples to
reach equilibration. A time-course measurement of water uptake and phase
transition during sorption was performed for samples stored in desiccators
containing saturated solutions of sodium chloride (75% RH) at a constant
7)
temperature of 25 °C.
X-Ray Powder Diffraction (XRPD) X-Ray powder diffraction (XRPD)
patterns were measured using a Rint Ultimaꢃ 2100 (Rigaku Corporation,
Tokyo, Japan) with CuKa radiation at 40 kV and 50 mA. Samples were
measured in steps of 0.02° with a scan speed of 6° per minute from 3° to
4
0°, 2q.
NMR Measurement A CMX-300 infinity spectrometer (Chemagnetics
Inc., CO, U.S.A.) was operated at a resonance frequency of 46.1 MHz. The
spin–lattice relaxation time (T ) of deuterium oxide absorbed in the formula-
1
tion was determined by an inversion recovery pulse sequence with 4096
pulse repetitions and a recycling time of 2.0 s.
Results and Discussion
Stability The stability of TAK-457 in lyophilized prod-
ucts was investigated under conditions of varying water con-
tent and temperature using two kinds of formulations: an
inositol formulation (IF) containing inositol as the excipient
and a mannitol formulation (MF) containing mannitol as the
excipient. Samples containing 2% and 8% water were pre-
pared by exposing the intact lyophilized formulations in glass
Fig. 2. HPLC Chromatograms Obtained from IF (A) and MF (B) Contain-
vials to moisture. The vials were sealed and then stored at ing 8% Water Stored at 25 °C
Fig. 3. The Chemical Stability of TAK-457 in Formulations Containing 2% Water (A) and 8% Water (B)
The percent of the initial amount (ꢄcontent at each time point/content at initial, %) of TAK-457 is shown by: ꢀ (IF, 25 °C), ꢁ (IF, 40 °C), ꢂ (MF, 25 °C), ꢃ (MF, 40 °C).

May 2009
461
Fig. 4. Isothermal Water-Sorption and -Desorption Profiles of IF (A) and MF (B) at 25 °C
Closed and open circles indicate the profiles under the conditions of increasing and decreasing humidity, respectively.
fluenced by water mobility rather than the amount of total
2—4)
water.
Although other parameters that could affect the
formulation stability should also be considered including the
molecular mobility of the drug itself, this report is focusing
on the association of water with the formulations and water
mobility in the two formulations, both of which could have
considerable influence on drug stability, as well as how varia-
tions in water contents result in changes to stability superior-
ity.
Hygroscopicity The sorption and desorption isotherms
between 5% and 95% relative humidity (RH) at 25 °C were
measured using a humidity-controlled microbalance system.
As shown in Fig. 4, although both formulations took up
moisture up to a maximum of about 13%, IF showed a
Fig. 5. Moisture Uptake Profiles of IF (ꢂ) and MF (ꢁ) under 75% RH at
25 °C
weight decrease in the course of the sorption process be- of the peaks observed in the stored IF samples (Fig. 6B) were
tween 60% RH and 65% RH, whereas MF showed a stable due to anhydrous inositol (Fig. 6C). Once the weight fell to
increase with humidity. A similar result was also reported for 8%, there was no further change in its weight or the diffrac-
a mixture of amorphous and crystalline lactose monohydrate tion pattern up to 24 h, meaning that the stored formulation
8)
by Buckton and Darcy, in which the amorphous region re- was composed of amorphous drug and anhydrous inositol.
crystallized during moisture uptake due to a plasticizer effect The intact MF sample showed some diffraction peaks (Fig.
that reduced the glass transition temperature of the amor- 7E). These peaks were consistent with those of Form-III
2
)
11)
phous materials. It was presumed that when moisture up- mannitol, as assigned by Burger et al., which is known to
11,12)
take reached about 10% in IF, the glass transition tempera- be generated by freeze-drying.
The stored MF samples
ture of the amorphous material was reduced to allow crystal- had no phase transition and an 8.5% moisture uptake for 24 h
lization at 25 °C and that this was accompanied with expul- (Fig. 7D). When the vials of the stored IF and MF were
sion of the absorbed water from the material, resulting in sealed and placed at 50 °C for 12 h, the crystallization of
1,8)
weight loss. The desorption profile of IF showed the con- amorphous drug to hydrate was observed in MF (Fig. 7C),
tinuous removal of adsorbed water as did that of MF.
but no further phase transition was observed in IF.
Phase Transition in Formulations with Moisture Up-
The results show that amorphous drug hydrated with
take The polymorphism of components in the formulations Form-III mannitol whereas no hydration occurred with anhy-
was determined during the course of moisture uptake. Intact drous inositol even though the total water contents were simi-
lyophilized formulations were stored in open glass vials lar in both formulations. This is probably caused by differ-
under a constant humidity of 75% RH at 25 °C using a desic- ences in the water affinity of the excipients. It was presumed
cator with a saturated solution of sodium chloride. The that amorphous inositol associated more strongly with water
weight change and XRPD patterns were measured over time than Form-III mannitol, leading to the superior chemical sta-
for 24 h.
bility of IF to MF in the presence of a low water content.
The moisture-uptake profile of IF (Fig. 5) was indicative However, if the phase transition of inositol to the anhydrate
of the crystallization of amorphous components as discussed occurs by water absorption, it should decrease the water
above. From the XRPD pattern, it was shown that both the affinity of the excipient. Additionally the released water
drug and inositol in the intact IF were of an amorphous state might have significant influence on the chemical stability of
(
(
Fig. 6A). From initiation to 4 h with a stable weight increase drug (discussed in detail later). These influences by the crys-
Fig. 5), no change was observed in the XRPD patterns of tallization of inositol were thought to lead to IF losing its su-
the IF samples, which showed no diffraction peaks. After the periority in the presence of a high water content.
weight started to decrease at 4 h, crystalline peaks appeared Molecular Mobility of Water Estimated by NMR It
Fig. 6B). It has been found that the drug substance (TAK- has been demonstrated in some reports that the stability of
(
4
57) can be crystalline as an anhydrate or a hydrate, and that moisture-sensitive drugs correlates well with the amount of
9)
10)
inositol can also be as an anhydrate or a dihydrate. From highly mobile water, a reactant of hydrolysis, rather than total
a comparison of the XRPD patterns with all possible forms water.
4
,5,13)
In a comparison of cephalothin stability among
for the drug and inositol (Figs. 6C—F) it was found that all mixtures of cephalothin and various excipients, it was sug-

4
62
Vol. 57, No. 5
2
Fig. 8. H-NMR Spectra of D O in IF (A) and MF (B) Along with Bulk
2
D O (C) at 20 °C, 40 °C, and 55 °C
2
As pointed out by arrows in (A), sharp signals were observed at 40 °C and 55 °C,
which were superimposed on top of broad signals.
Fig. 6. XRPD Patterns of IF and Its Components
(
A) IF, intact; (B) IF, unsealed and stored at 25 °C/75% RH for 5 h; (C) inositol, an-
hydrate; (D) inositol, dihydrate*; (E) TAK-457, anhydrate; and (F) TAK-457, hydrate.
The peaks indicated by arrows represent characteristic peaks of the anhydrous drug (E).
Calculated using crystal data by Mercury.
2
Table 1. H Spin–Lattice Relaxation Time (T ) of Deuterium Oxide in the
1
Formulations with 7—8% Water and Bulk Water
10)
17)
*
T1 (ms)
Temperature
a)
b)
IF
MF
Bulk water
20 °C
40 °C
55 °C
5.3
6.3, 27.6
7.7, 114.6
5.3
7.0
8.5
270.2
410.5
516.6
c)
d)
c)
d)
a) With 8% water, b) with 7% water, c) calculated from a broader peak, d) calculated
from a narrower peak.
terium oxide in the formulations were measured at 20 °C,
4
0 °C, and 55 °C (Fig. 8, Table 1).
It is recognized that a broader signal may be derived from
Fig. 7. XRPD Patterns of Stored MF and TAK-457
water molecules with lower mobility, although the contribu-
(A) TAK-457, anhydrate; (B) TAK-457, hydrate; (C) MF, sealed and stored at 50 °C
2
tion of the chemical exchange of H between deuterium
for 12 h following the moisture-uptake experiment; (D) MF, unsealed and stored at
2
5 °C/75% RH for 24 h; and (E) MF, intact. The peaks indicated by arrows represent oxide and a hydroxyl group of the excipients to the signal
characteristic peaks of the anhydrous drug (A). The XRPD pattern (E) was consistent
5)
broadening cannot be excluded in this study. In Fig. 8 broad
signals were observed from IF and MF compared with bulk
water indicating that water mobility was restricted in each of
1
1)
with that of Form-III mannitol.
4
,5)
gested by Aso et al. that water mobility estimated by the the formulations. MF showed a single broad peak at each
2
H-NMR technique could be used as a measure to investigate temperature and the peak became narrower due to the in-
differences in drug stability. Water mobility in the two model crease of water mobility as the temperature was elevated. On
2
formulations of TAK-457 was measured by H-NMR in the other hand, in IF, there was an extremely broad single
order to estimate the association of water in the formulations. peak at 20 °C and a sharp peak superimposed on the broad
IF and MF with 7—8% deuterium oxide were prepared by peak was observed at 40 °C and 55 °C (Fig. 8A), indicating
placing intact formulations under 51% RH in a desiccator that there were at least two water populations with distinctly
containing a saturated deuterium oxide solution of calcium different mobilities above 40 °C. It is thought that the broad
nitrate. In this preparation, care was taken not to cause the and the sharp signals in IF were derived from water associat-
2
crystallization of inositol by moisture uptake. The H-NMR ing with inositol and released water generated by the crystal-
spectra and the spin–lattice relaxation times (T ) of deu- lization of inositol during the temperature increase. Even
1

May 2009
463
better chemical stability of the drug in IF. This shows that
utilization of suitable amorphous materials as excipients
could be a promising approach to stabilizing lyophilized for-
mulations of moisture-sensitive drugs. It should be noted,
however, that such amorphous materials might carry a risk of
causing crystallization by water absorption as exhibited in
this study. Therefore when such amorphous excipients are
used for lyophilized formulations, it is crucial to evaluate the
amount of water which formulations can hold without risk of
crystallization. It is also important to protect the formula-
tions from moisture as much as possible and to monitor
2
Fig. 9. H-NMR Spectra of D O in IF in the Presence of 2% Water (A)
2
and 8% Water (B) at 40 °C
water content adequately during the manufacturing process
2
though signal was not detected from IF at 20 °C in the pres- and storage. This study also indicated that H-NMR analysis
ence of a low water content (2%) due to insufficient peak in- of water mobility is a useful tool for investigating the chemi-
tensity, a broad single peak was observed by increasing tem- cal and physical stability of drugs in formulations.
perature up to 40 °C, indicating that only associated water
Acknowledgements The authors would like to thank Mr. Masao Nagao
for assistance with the preparation of the lyophilized formulations and Toray
Research Center for performing the NMR measurement and for helpful dis-
existed in the formulation (Fig. 9). Because the signal of re-
stricted water in IF was broader than that in MF at each tem-
perature, it was suggested that IF had a water population with cussion.
a lower mobility than that of MF due to the higher water
References
affinity of inositol than Form-III mannitol.
1)
Brittain H. G., “Polymorphism in Pharmaceutical Solids,” Marcel
Dekker Inc., New York, 1999, pp. 408—410.
The spin–lattice relaxation time (T ) presented more quan-
1
5)
titative information. At 40 °C and 55 °C, two T s correspond-
1
2) Ahlneck C., Zografi G., Int. J. Pharm., 62, 87—95 (1990).
3) Herman B. D., Sinclair B. D., Milton N., Nail S. L., Pharm. Res., 11,
1467—1473 (1994).
ing to the broad signals (the associated water) and the sharp
signals (the released water) were calculated for IF. As shown
4)
Aso Y., Sufang T., Yoshioka S., Kojima S., Drug Stability, 1, 237—
in Table 1, the T value for each water population increased
1
1
4—16)
242 (1997).
with temperature due to the increase of mobility.
The
5)
Aso Y., Yoshioka S., Terao T., Chem. Pharm. Bull., 42, 398—401
much shorter values of T for the broad signals of the associ-
1
(1994).
ated water in IF and MF than the value of bulk water showed
that the water mobility was largely restricted in each of the
6) Ichikawa T., Kitazaki T., Matsushita Y., Yamada M., Hayashi R.,
Yamaguchi M., Kiyota Y., Okonogi K., Itoh K., Chem. Pharm. Bull.,
4
9, 1102—1109 (2001).
formulations. Although the longer T values for the sharp
1
7)
8)
9)
Greenspan L., J. Res. Natl. Bur. Stand., 81A, 89—96 (1977).
Buckton G., Darcy P., Int. J. Pharm., 123, 265—271 (1995).
Rabinowitz I. N., Kraut J., Acta Crystallogr., 17, 159—168 (1964).
signals observed above 40 °C in IF than the values corre-
sponding to the associated water represented the released
water with high mobility, these values also indicated obvious 10) Lomer T. R., Miller A., Beevers C. A., Acta Crystallogr., 16, 264—
2
68 (1963).
interactions between the released water and the formulation
from comparison with still longer values of bulk water.
1
1) Burger A., Henck J. O., Hetz S., Rollinger J. M., Weissnicht A. A.,
Stöttner H., J. Pharm. Sci., 89, 457—468 (2000).
1
2) Izutsu K., Yoshioka S., Terao T., Chem. Pharm. Bull., 42, 5—8 (1994).
Conclusions
13) Heidemann D. R., Jarosz P. J., Pharm. Res., 8, 292—297 (1991).
It was concluded that amorphous inositol associated more 14) Yoshioka S., Aso Y., Kojima S., J. Pharm. Sci., 91, 2203—2210
(
2002).
5) Bloembergen N., Purcell E. M., Pound R. V., Phys. Rev., 73, 679—712
1948).
strongly with water, a chemical reactant of drug hydrolysis,
in a lyophilized formulation than Form-III mannitol and that
water would be less reactive with the drug in IF than in MF.
The NMR measurement suggested that water motion was
1
1
(
6) Andrew E. R., Hinshaw W. S., Hutchins M. G., Sjöblom R. O. I., Mol.
Phys., 31, 1479—1488 (1976).
more restricted in IF than in MF, being consistent with the 17) Mercury 1.2.1., ꢀhttp://www.ccdc.cam.ac.uk/mercury/ꢁ, CCDC.