indomethacin

European Journal of Pharmaceutics and Biopharmaceutics 109 (2016) 140–148

Contents lists available at ScienceDirect

European Journal of Pharmaceutics and Biopharmaceutics journal homepage: www.elsevier.com/locate/ejpb

Research paper

Influence of the cooling rate and the blend ratio on the physical stability of co-amorphous naproxen/indomethacin Andreas Beyer a, Holger Grohganz b, Korbinian Löbmann b, Thomas Rades b, Claudia S. Leopold a,⇑ a b

Division of Pharmaceutical Technology, University of Hamburg, Bundesstrabe 45, 20146 Hamburg, Germany Department of Pharmacy, University of Copenhagen, Universitetsparken 2, DK-2100, Denmark

a r t i c l e

i n f o

Article history: Received 26 June 2016 Revised 10 October 2016 Accepted in revised form 12 October 2016 Available online 13 October 2016 Keywords: Co-amorphous Binary Cooling Composition Eutectic Equimolar Physical stability

a b s t r a c t Co-amorphization represents a promising approach to increase the physical stability and dissolution rate of amorphous active pharmaceutical ingredients (APIs) as an alternative to polymer glass solutions. For amorphous and co-amorphous systems, it is reported that the preparation method and the blend ratio play major roles with regard to the resulting physical stability. Therefore, in the present study, co-amorphous naproxen-indomethacin (NAP/IND) was prepared by melt-quenching at three different cooling rates and at ten different NAP/IND blend ratios. The samples were analyzed using XRPD and FTIR, both directly after preparation and during storage to investigate their physical stabilities. All cooling methods led to fully amorphous samples, but with significantly different physical stabilities. Samples prepared by fast cooling had a higher degree of crystallinity after 300 d of storage than samples prepared by intermediate cooling and slow cooling. Intermediate cooling was subsequently used to prepare co-amorphous NAP/IND at different blend ratios. In a previous study, it was postulated that the equimolar (0.5:0.5) co-amorphous blend of NAP/IND is most stable. However, in the present study the physically most stable blend was found for a NAP/IND ratio of 0.6:0.4, which also represents the eutectic composition of the crystalline NAP/c-IND system. This indicates that the eutectic point may be of major importance for the stability of binary co-amorphous systems. Slight deviations from the optimal naproxen molar fraction led to significant recrystallization during storage. Either naproxen or c-indomethacin recrystallized until a naproxen molar fraction of about 0.6 in the residual co-amorphous phase was reached again. In conclusion, the physical stability of co-amorphous NAP/IND may be significantly improved, if suitable preparation conditions and the optimal phase composition are chosen. Ó 2016 Published by Elsevier B.V.

1. Introduction The majority of active pharmaceutical ingredients (APIs) is administered as solid dosage forms and reaches the blood stream by the oral route. This concept is convenient and cost effective, but has its limitations, as the API has to dissolve in the aqueous

Abbreviations: cNAP, crystalline naproxen; DSC, differential scanning calorimetry; Fam, amorphous fraction; FC, fast cooling; FTIR, Fourier-transformed infrared spectroscopy; IBragg, intensity under the Bragg peaks; IC, intermediate cooling; IHalo, intensity under the halo baseline; IND, indomethacin; NMF, naproxen molar fraction; NAP, naproxen; NAP/IND, naproxen-indomethacin; P2O5, phosphorous pentoxide; PC, principal component; PCA, principal component analysis; RIR, relative intensity ratio; SC, slow cooling; XRPD, X-ray powder diffractometry. ⇑ Corresponding author. E-mail addresses: [email protected] (A. Beyer), [email protected] (H. Grohganz), [email protected] (K. Löbmann), [email protected] (T. Rades), claudia.leopold@ pharmaceutical-technology.de (C.S. Leopold). http://dx.doi.org/10.1016/j.ejpb.2016.10.002 0939-6411/Ó 2016 Published by Elsevier B.V.

digestive media and to diffuse through the intestinal membrane [1]. Therefore, sufficient water solubility and membrane permeability are mandatory for APIs to become orally bioavailable. However, many of the new potential drug candidates are poorly watersoluble while they are acceptably permeable and therefore belong to class II of the Biopharmaceutics Classification System [2,3]. For these drugs, an increase of the solubility and thus the dissolution rate is required to improve their oral bioavailability [4,5]. Amorphous solids can be prepared by destruction of the integrity of the crystal lattice of a crystalline API by solution, melt or mechanical activation approaches [6]. As no long range order exists in amorphous solids, a lower energy barrier has to be overcome to dissolve the molecules. However, because of the higher internal energy of solids in the amorphous form compared to their respective crystalline counterparts, amorphous systems are physically unstable and tend to recrystallize during storage [7,8]. Furthermore, amorphous phases are sensitive to mechanical stress [9],

A. Beyer et al. / European Journal of Pharmaceutics and Biopharmaceutics 109 (2016) 140–148

thermal stress [10] as well as solvents [11], i.e. aspects which are important with regard to their manufacturability and processability. To improve the physical stability of amorphous phases, stabilizing agents such as pharmaceutical polymers are often used to prepare solid polymer solutions [1,12,13]. However, this approach is often accompanied by problems regarding the manufacturability [8,14,15], the hygroscopicity [16] of many polymers and the often limited miscibility of the APIs with polymers [1]. As an alternative to pharmaceutical polymers, low molecular weight molecules may be added as stabilizers to form coamorphous systems, in which the involved molecules are mixed on a molecular level to form one single co-amorphous phase [17]. Co-amorphous systems comprising two excipients [18], two suitable APIs [19] or API plus excipient [20,21] have already been described in the literature. In these systems, molecular interactions [22,23] between the involved components play a major role. In coamorphous naproxen/cimetidine and ranitidine-HCl/ indomethacin, 1:1 interactions between the two compounds were observed [19,24]. Similarly, with co-amorphous naproxen/indomethacin (NAP/IND) the formation of heterodimers has been shown [25,22]. The ‘‘1:1 theory” is supported by previous studies reporting the highest physical stability for the respective equimolar co-amorphous systems [19,24,25]. The physical stability of the non-equimolar co-amorphous 1:2 and 2:1 blends was observed to be decreased and it was found that the respective excess components recrystallized first [19,24,25]. Therefore, specific interactions between the involved compounds in the co-amorphous phase have been suggested to be of major importance for the physical stability and recrystallization behavior of these systems [19,24,25,20]. In the present study, co-amorphous NAP/IND was chosen as a model system to examine its recrystallization behavior as a function of the molar ratio of NAP and IND in the co-amorphous phase in more detail to further evaluate the ‘‘1:1 theory” for co-amorphous binary systems. For single-amorphous systems it was reported that the physical stability is dependent on the cooling rate during melt-quenching and that a faster cooling procedure results in more stable samples [10]. Therefore, the influence of the cooling rate during preparation of co-amorphous NAP/IND [25] by melt-quenching on the resulting physical stability of the system was also investigated. To quantify the molar fractions of naproxen and c-indomethacin that recrystallized during storage, the respective X-ray powder diffractograms were evaluated with the relative intensity ratio method. 2. Materials and methods 2.1. Materials Crystalline naproxen (cNAP, Mr = 230.26 g/mol, Tm = 158 °C) and c-indomethacin (c-IND, Mr = 357.79 g/mol; Tm = 162 °C) were purchased from Fagron (Barsbüttel, Germany) and used as received. 2.1.1. Preparation of equimolar co-amorphous naproxenindomethacin with different cooling rates An amount of 1 g of equimolar co-amorphous naproxen/indomethacin (NAP/IND) was prepared in triplicate by melting the equimolar physical mixture of cNAP and c-IND at 170 °C for 10 min and applying three different cooling methods: Fast cooling (FC): Liquid nitrogen (N2) was poured onto the melts, which were subsequently transferred into a P2O5 desiccator. After the N2 was fully evaporated while displacing the present air through a hole to avoid moisture sorption, the hole was closed until the samples reached room temperature again. Intermediate cooling (IC): Within a P2O5 desiccator, the melts cooled down to room

141

temperature. Slow cooling (SC): The samples were covered with aluminum foil, transferred into a HAAKE C25P thermostat (Thermo Fisher, Waltham, USA) and cooled down to room temperature within 180 min. 2.1.2. Preparation of co-amorphous naproxen/indomethacin at different molar ratios 1 g of each of the respective physical mixtures was molten at 170 °C for 10 min and subsequently cooled down to room temperature inside a desiccator above P2O5 using the IC approach as described above. Ten different NAP/IND blends with the following naproxen molar fractions (NMFs) were prepared in triplicate: 0.1, 0.2, 0.3, 0.4, 0.5, 0.55, 0.6, 0.7, 0.8, 0.9. 2.1.3. Sample processing and storage The differently prepared samples were homogenized, particle size-reduced and sieved (250 lm) in an air conditioned room at 6 °C by mortar and pestle before they were stored in open sample tubes within desiccators (21 °C, P2O5). The differently cooled samples were stored for up to 300 days while sampling was performed after 0, 56, 112 and 300 days. The samples at different naproxen molar fractions were stored for up to 112 days and analyzed after 0, 56 and 112 days. 2.2. Methods 2.2.1. X-ray powder diffractometry (XRPD) An X’Pert PRO X-ray diffractometer (PANalytical, Almelo, The Netherlands; Cu Ka anode; k = 1.5406 Å; 45 kV; 40 mA) was used to record X-ray powder diffractograms between 5 and 35° 2h with a scan speed of 0.045° 2h/min and a step size of 0.0131° 2h. The obtained diffractograms were separately baseline offset-corrected and normalized to unit area [26] using The Unscrambler X software (ver. 10.3, CAMO Software, Oslo, Norway). Savitzky-Golay smoothing was performed using The Unscrambler X software (polynomial order: 1; 9 smoothing points) for better visualization. 2.2.2. Relative intensity ratio (RIR) In order to quantify the total amorphous fraction (Fam) of the prepared samples directly after preparation and during storage, the relative intensity ratio (RIR) method was used to evaluate the XRPD data [27]. Briefly, the background determination function of the Highscore Plus software (ver. 2.2, PANalytical, Almelo, The Netherlands) was used to quantify the halo intensities (IHalo), resulting from the amorphous portion of the samples and the Bragg peak intensities (IBragg), which emerge from the crystalline portion of the samples in each diffractogram. Subsequently, Eq. (1) was used to calculate the relative intensity ratio of the halo (RIRHalo):

RIRHalo ¼

IHalo
IHalo þ IBragg

ð1Þ

Based on the resulting values for RIRHalo and the calibration function in Eq. (2), which was set up in a former study [28], the total amorphous fraction of each sample was calculated:

Fam ¼ 150  RIRHalo  35:6

ð2Þ

2.2.3. Fourier transformed infrared (FTIR) spectroscopy A Tensor 37 spectrometer (Bruker Optik GmbH, Ettlingen, Germany) equipped with a MIRacle attenuated total reflectance device with diamond crystal plate (Piketech, Madison, USA) and the Opus software ver. 7 was used to record infrared spectra over a range of 4000–400 cm1 (128 scans, resolution 4 cm1). The spectral region between 1600 and 1800 cm1 was selected [25] for further analysis, after baseline offset-correction as well

142

A. Beyer et al. / European Journal of Pharmaceutics and Biopharmaceutics 109 (2016) 140–148

as normalization to unit area [26] was performed using The Unscrambler X software. 2.2.4. Differential scanning calorimetry (DSC) DSC thermograms were recorded with a DSC 1 (Mettler Toledo, Columbus, USA). For each sample, an amount of 10–20 mg was weighed into aluminum pans and closed with pierced lids. Samples were scanned within a temperature range of 50 to 180 °C at a heating rate of 10 °C/min. Data were analyzed using STARe Software (Mettler Toledo, Columbus, USA). 2.2.5. Karl-Fischer analysis Determination of the water content was carried out by coulometric Karl-Fischer analysis with 20–30 mg of each sample with a Mettler Toledo DL37 KF Coulometer (Mettler Toledo, Columbus, USA). 3. Results and discussion 3.1. Influence of the cooling rate on the physical stability of coamorphous NAP/IND To study the influence of the cooling rate on the physical stability of equimolar NAP/IND, three different cooling rates (FC, IC, SC) were investigated. Karl Fischer and DSC analysis were performed on the freshly prepared samples and revealed similar water contents and glass transition temperatures (Table 1), which were both not significantly different (p > 0.05). The obtained XRPD data were evaluated by principal component analysis (PCA). In Fig. 1A, the corresponding mean PC-1 scores ± SD (n = 3) are plotted versus the storage time periods of the differently prepared samples. For all storage time periods, the PC-1 scores of the FC samples and the corresponding standard deviations are higher compared to the IC and SC samples, especially after 300 days of storage. To interpret the PC-1 scores, the PC-1 loadings plot together with the reference diffractogram of c-IND and cNAP is shown in Fig. 1B. All peak signals in the positive part of the PC-1 loading are assignable to c-IND or cNAP, while no peak signals but a halo shape is visible in the negative part of the PC-1 loading. Samples with high (positive) PC-1 scores show diffractograms with distinct signals according to the positive part of the PC-1 loading: recrystallization peaks of c-IND and cNAP. Samples with low (negative) PC-1 scores show diffractograms similar to the negative part of the PC-1 loading: diffractograms with a halo signal. Therefore, the higher the PC-1 scores of the respective samples, the more distinct are the recrystallization peaks of c-IND or cNAP. The lower the PC-1 scores, the more amorphous are the samples. Obviously, the FC samples have the highest PC-1 scores and therefore recrystallized faster than the SC samples, while the IC samples are the least crystalline. In Fig. 1C, one exemplary Savitzky-Golay smoothed diffractogram for each cooling method after 300 days of storage is shown for comparison. It is obvious, that the diffractogram of the IC samples includes less distinct diffraction peaks compared to the diffractograms of the SC and FC samples, which confirms the suggestion of the PCA: the IC samples are the least crystalline after 300 d of storage. However, the SC samples show some small additional peaks in the diffractogram, that can be Table 1 Mean water contents and glass transition temperatures ± SD (n = 3) of unstored FC, IC and SC samples. Samples

FC

IC

SC

Water content (%) Glass transition temperature (°C)

0.37 ± 0.16 31.5 ± 1.3

0.78 ± 0.14 26.7 ± 0.6

0.62 ± 0.28 30.7 ± 4.7

assigned to the a-IND polymorph small (the gray dotted lines), which may also explain why the SC samples recrystallized faster than the IC samples. To verify the results obtained from the XRPD dataset, FTIR analysis was performed, and in Fig. 2, one exemplary spectrum of the differently cooled samples, stored for 300 days, is shown. The peaks at 1689 cm1 (benzoyl function in c-IND), 1714 cm1 (bound COOH in c-IND) and 1725 cm1 (free COOHs in cNAP) are most pronounced with the FC samples and less distinct for the IC and SC samples. Therefore, the FTIR data confirm the results of the XRPD data: the FC samples are physically less stable than the IC and SC samples and recrystallization proceeds under formation of c-IND and cNAP. To quantify the total crystalline fraction Fcryst of the differently cooled samples during storage, the RIR method (described in Section 2.2.2.) was used to evaluate the XRPD data. In Fig. 3, the mean total crystalline fractions Fcryst ± SD (n = 3) of the samples are plotted versus the storage time periods and show that FC leads to samples that are 0.48 ± 0.28 crystalline after 300 d of storage, in contrast to 0.15 ± 0.03 (IC samples) and 0.20 ± 0.01 (SC samples). These results are in good agreement with the PCA data presented in Fig. 1A and confirm that the temperature program plays a major role for the physical stability of co-amorphous NAP/IND. Interestingly, FC leads to a significantly less stable product, which in addition recrystallized in a less reproducible manner. This recrystallization behavior is probably caused by mechanical stress that is applied to the product during the fast temperature decrease of the samples from 170 °C to 196 °C within seconds [9] and by the induction of cracks and heterogeneity in the material, which promotes recrystallization of the FC samples [29,30]. Moreover, SC only slightly increases the recrystallization rate of co-amorphous NAP/IND compared to IC. This finding is surprising, as it has previously been reported that the preparation of amorphous materials by slow cooling leads to physically less stable materials, e.g. with plain amorphous indomethacin [10]. Therefore, the optimal cooling rate for the preparation of co-amorphous NAP/IND is neither maximal nor minimal but intermediate. 3.2. Influence of the blend ratio on the physical stability of coamorphous NAP/IND Because IC led to the physically most stable samples after 300 days of storage compared to SC and FC, IC was also performed in the second part of the study. Here it was the aim to investigate the influence of the phase composition on the resulting physical stability of the co-amorphous system. Therefore, co-amorphous NAP/IND was prepared in ten different blend ratios by IC. Subsequently, the samples were stored and characterized by XRPD and FTIR to evaluate their physical stabilities. 3.2.1. Unstored samples In Fig. 4A, one exemplary diffractogram is shown for each of the unstored samples with increasing NMFs ranging from 0.1 to 0.7. All diffractograms comprise exclusively halo signals and thus, the samples are fully amorphous. In contrast, the 0.8 and 0.9 samples (Fig. 4B) show distinct peak intensities for cNAP, revealing that these samples already recrystallized to a significant level. This is in accordance with observations made during sample preparation in both samples, i.e. a crystalline phase was formed at 170 °C after the samples were initially fully molten. In Fig. 5A, one exemplary FTIR spectrum of each of the unstored samples with NMFs between 0.1 and 0.7 is shown. The bands around 1735 cm1 (free COOH groups), 1701–1705 cm1 (NAP/IND interaction) and 1678–1684 cm1 (benzoyl function in IND) represent the different carbonyl functions present in co-amorphous NAP/IND [25]. The shift of the NAP/IND interaction band from 1705 to

143

A. Beyer et al. / European Journal of Pharmaceutics and Biopharmaceutics 109 (2016) 140–148

0.018 0.012 FC

0.006

IC

0.000 SC

0

56

112

FC

C

-0.006

IC SC

300

Intensity

PC-1 scores (84%)

A

Storage time (d)

B PC-1

5

10

15

20

25

30

35

PC loadings and intensities

Angle/2 (°) -IND -IND

5

10

15

20

25

30

35

Angle/2 (°) cNAP

5

10

15

20

25

30

35

Angle/2 (°)

Absorption

cNAP

A

-IND

Intensity

Fig. 1. (A) Mean PC-1 scores ± SD (n = 3) of the differently cooled NAP/IND samples plotted versus the storage time. (B) Corresponding PC-1 loadings plot as well as reference diffractograms of c-IND and cNAP. (C) Exemplary Savitzky-Golay smoothed diffractograms of the FC, IC and SC samples, stored for 300 d, and for a-IND.

5

REF

10

15

20

25

30

35

Angle/2 (°) 0.1 0.5

0.2 0.55

0.3 0.6

0.4 0.7

B 1750

1700

1650

1600

Intensity

1800

(cm-1) FC

IC

SC

Fig. 2. Exemplary FTIR spectra of the FC, IC and SC samples, stored for 300 d, and reference spectra of cNAP, c-IND and unstored equimolar co-amorphous NAP/IND (REF).

cNAP

5

10

15

20

0.8

30

35

0.9

Fig. 4. (A) Exemplary diffractogram of each of the unstored samples with NMFs between 0.10 and 0.70. (B) Exemplary diffractogram of the unstored samples with NMFs of 0.80 and 0.90 and reference diffractogram of cNAP.

1.0 0.8

Fcryst

25

Angle/2 (°)

FC

0.6

IC

0.4

SC

0.2 0.0 0

56

112

300

Storage time (d) Fig. 3. Mean total crystalline fractions (FCryst) ± SD (n = 3) of the FC, IC and SC samples plotted versus the storage time.

1701 cm1 with increasing NMF indicates that the location of the peak depends on the NMF in the co-amorphous phase. Furthermore, with increasing NMF, the IND benzoyl band at 1678 cm1 [31] decreases and also slightly shifts to higher wave numbers of 1684 cm1, which indicates that the benzoyl group is also involved in the interaction between NAP and IND. In contrast, the exemplary FTIR spectra of the 0.8 and 0.9 samples in Fig. 5B show a distinct decrease of the NAP/IND interaction band (1701–1705 cm1) and

144

A. Beyer et al. / European Journal of Pharmaceutics and Biopharmaceutics 109 (2016) 140–148

A

Table 2 Mean water contents ± SD (n = 3) of the unstored samples with NMFs between 0.1 and 0.7.

NAP/IND interaction

Absorption

Benzoyl function Free COOH

1800

1750

1700

1650

1600

(cm-1) 0.2 0.55

Absorption

B

0.3 0.6

Free COOH

cNAP

1800

Benzoyl function

1700

1650

1600

(cm-1) 0.8

0.72 ± 0.17 0.91 ± 0.11 1.10 ± 0.13 1.10 ± 0.18 0.90 ± 0.17 0.85 ± 0.09 0.82 ± 0.03 0.77 ± 0.11

3.2.2. Stored samples After 56 and 112 days of storage, all prepared samples were analyzed by XRPD and FTIR to assess their physical stability. All resulting diffractograms and spectra were evaluated by PCA. According to the resulting PCA scores plot for the XRPD dataset (Fig. 7A), all stored samples with NMFs between 0.1 and 0.6 are mainly described by PC-1 while increasing PC-1 scores correlate with decreasing NMFs. In contrast, the 0.7 samples are mainly described by PC-2.

Bound COOH

NAP/IND interaction Free COOH

1750

0.4 0.7

Water content (%)

0.1 0.2 0.3 0.4 0.5 0.55 0.6 0.7

0.9

Fig. 5. (A) Exemplary FTIR spectrum of each of the unstored samples with NMFs between 0.10 and 0.70. (B) Exemplary spectrum of the unstored samples with NMFs of 0.80 and 0.90 and reference diffractogram of cNAP.

A

0.01

PC-2 (10%)

0.1 0.5

NMF

an increase at 1725 cm1 (free COOHs in cNAP) and 1680 cm1 (bound COOHs in cNAP), which both represent absorption bands present in cNAP [25]. Therefore, the XRPD and FTIR data are consistent: co-amorphous NAP/IND is formed with NMFs between 0.1 and 0.7, while the samples with NMFs of 0.8 and 0.9 recrystallized under formation of only cNAP. DSC reveals only single-phase glass transition temperatures of the 0.1–0.7 samples (Fig. 6) comparable to those reported by Löbmann et al. [25], which indicates that single-phase coamorphous systems were obtained. Moreover, the samples only contain low but comparable amounts of water (Table 2) and therefore, the influence of the water content on the physical stability of the samples is expected to be negligible.

-0.01

0.02

-0.01 0.10 56d 0.30 56d 0.50 56d 0.60 56d unstored

PC-1 (80%)

0.10 112d 0.30 112d 0.50 112d 0.60 112d

0.20 56d 0.40 56d 0.55 56d 0.70 56d

0.20 112d 0.40 112d 0.55 112d 0.70 112d

B PC loadings and intensities

50

Tg (°C)

40 30 20 10 0.00

-IND

0.10

0.20

0.30

0.40

0.50

0.60

PC-1

cNAP

PC-2

0.70

NMF Current study

Löbmann et al.

5

10

15

20

25

30

35

Angle/2 (°) Fig. 6. Mean glass transition temperatures ± SD (n = 3) of the unstored initially fully amorphous samples plotted versus the naproxen molar fraction (NMF). For comparison, the mean glass transition temperatures reported by Löbmann et al. are shown.

Fig. 7. (A) PCA scores plot for the XRPD dataset of the stored NAP/IND samples with NMFs between 0.10 and 0.70. (B) Corresponding PC-1 and PC-2 loadings plots as well as reference diffractograms of c-IND and cNAP.

145

A. Beyer et al. / European Journal of Pharmaceutics and Biopharmaceutics 109 (2016) 140–148

FcIND ¼ 1-Fam

ð3Þ

FcNAP ¼ 1-Fam

ð4Þ

In Fig. 10A, the calculated fractions Fc-IND and FcNAP present in the stored samples are plotted versus their NMFs. The V-like shape of the data indicates that the 112 days-physical stability is strongly

A PC-2 (30 %)

0.002

-0.003

0.003

-0.002

PC-1 (52 %) 0.10 0d

0.10 56d

0.10 112d

0.20 0d

0.20 6d

0.20 112d

0.30 0d

0.30 56d

0.30 112d

0.40 0d

0.40 56d

0.40 112d

0.50 0d

0.50 56d

0.50 112d

0.55 0d

0.55 56d

0.55 112d

0.60 0d

0.60 56d

0.60 112d

0.70 0d

0.70 56d

0.70 112d

0.80 0d

0.80 56d

0.80 112d

0.90 0d

0.90 56d

0.90 112d

B Absorption and PC loadings

In the positive part of the PC-1 loadings plot (Fig. 7B), exclusively diffraction signals according to the c-IND reference diffractogram are found, while in the negative part predominantly halo signals are present. Therefore, samples with high (positive) PC-1 scores show diffraction peaks similar to c-IND, while samples with low (negative) PC-1 scores are predominantly amorphous. Thus, PC-1 separates the samples according to their c-IND crystallinity (high PC-1 scores) and amorphicity (low PC-1 scores). In the positive part of the PC-2 loadings plot predominantly diffraction intensities of cNAP are found, while a halo like shape is visible in the negative part. Therefore, samples with high (positive) PC-2 scores show diffraction peaks similar to cNAP, while samples with low (negative) PC-2 scores are predominantly amorphous. Thus, the XRPD data reveal that all samples with NMFs between 0.1 and 0.6 recrystallize predominantly under formation of c-IND, while the 0.7 samples recrystallize under formation of cNAP and the 0.50, 0.55 and 0.60 samples stay mostly amorphous. It appears, that the most stable samples after 56 and 112 days of storage are those with an NMF of 0.6, as these cluster in the lower left corner of the PCA scores plot (Fig. 7A), which is in contrast to literature reports postulating that the equimolar naproxenindomethacin system (with a NMF of 0.5) represents the physically most stable composition for co-amorphous NAP/IND [25] and for binary co-amorphous systems in general [19,24]. To confirm the results from the XRPD dataset, FTIR analysis was performed, and the obtained spectra were also evaluated by PCA. The resulting PCA scores plot (Fig. 8A) divides the stored samples into three groups predominantly along the PC-1 axis. In the negative part of the PC-1 axis, the 0.9 and 0.8 samples can be found, while in the positive part all samples with NMFs from 0.40 to 0.10 are located. In contrast, the 112 days stored samples with NMFs of 0.7, 0.6, 0.55 and 0.5 cluster near the center of the scores plot (Fig. 8A). As the PC-1 loading (Fig. 8B) shows c-IND signals in the positive and cNAP signals in the negative part, the samples with NMFs of 0.8 and 0.9 (in the negative part of PC-1) recrystallize under formation of cNAP while those with NMFs of 0.4–0.1 (in the positive part of PC-1) recrystallize predominantly under formation of c-IND, which confirms the results of the XRPD dataset. The PC-2 loadings plot (Fig. 8B) shows cNAP and c-IND FTIR signals in the positive and REF signals in the negative part. Therefore, PC-2 separates crystalline samples (in the positive part of PC-2) from coamorphous samples and the stored 0.6 samples that cluster at PC-1 scores near the zero line and at low PC-2 scores are the least crystalline, as already shown by XRPD. The PCAs of the XRPD and FTIR dataset indicate that either cNAP or c-IND is formed during storage depending on the NAP/IND blend ratio in the co-amorphous phase. To confirm these observations, the diffractograms of the samples with NMFs between 0.1 and 0.6 as well as those with NMFs between 0.7 and 0.9 are shown together with the reference diffractograms of c-IND and cNAP in Fig. 9. It is obvious, that the peaks of the first group can predominantly be assigned to c-IND while the peaks of the second group are predominantly assignable to the reference diffractogram of cNAP. Therefore, the PCA-suggested formation of either c-IND or cNAP is also found in the raw diffractograms. Knowing which phase predominantly recrystallizes in the differently composed samples, the total c-IND fraction Fc-IND and the total cNAP fraction FcNAP in the stored samples may be calculated based on the total amorphous fraction Fam according to Eqs. (3) and (4), respectively:

γ-IND

PC-1

cNAP PC-2

REF

1800

1750

1700

1650

1600

λ (cm-1) Fig. 8. (A) PCA scores plot for the FTIR dataset of the stored samples with NMFs between 0.10 and 0.90. (B) Corresponding PC-1 and PC-2 loadings plots as well as reference spectra of c-IND, cNAP and unstored equimolar co-amorphous NAP/IND (REF).

dependent on the NMF and that the optimal physical stability is found for the 0.6 (Fam = 0.94) followed by the samples with NMFs of 0.55 and 0.5 (Fam of about 0.89). However, slight deviations from the 0.5–0.6 NMF range lead to significant sample recrystallization. Based on the calculated Fc-IND and FcNAP fractions, the composition of the residual amorphous phase after storage in terms of the residual molar amorphous naproxen fraction (NMFResidual), after recrystallization occured, may be calculated by Eq. (5):

NMFResidual ¼ ðNMFInitial  FcNAP Þ=Fam

ð5Þ

In Fig. 10B, the residual NMFs after 56 and 112 days of storage are plotted. Interestingly, after 112 days of storage, the coamorphous phase of the samples with NMFs between 0.2 and 0.8 reaches residual NMFs near 0.6 (0.603 ± 0.047), indicating that almost all samples recrystallize until a NAP/IND ratio in the coamorphous phase near the previously found optimal composition with an NMF of 0.6 is reached again. To find an explanation for the increased physical stability of samples with NMFs between 0.5 and 0.6, the reported phase diagram of the c-IND/cNAP system by Löbmann et al. [25] was evalu-

146

A. Beyer et al. / European Journal of Pharmaceutics and Biopharmaceutics 109 (2016) 140–148

0.10

0.20

0.30 0.40

Intensities

0.50

0.55 0.60

-IND

cNAP

0.70 0.80

0.90

5

10

15

20

25

30

35

Angle/2 (°) Fig. 9. Identically scaled baseline offset corrected and normalized X-ray powder diffractograms of 112 d stored samples with NMFs between 0.10 and 0.90 versus the reference diffractograms of cNAP and c-IND.

B 1.0

1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

-IND

cNAP

NMFresidual

FcNAP or F IND

A

0.10 0.20 0.30 0.40 0.50 0.55 0.60 0.70 0.80 0.90

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.10 0.20 0.30 0.40 0.50 0.55 0.60 0.70 0.80 0.90

NMF

NMF 56 d

112 d

0d

56 d

112 d

Fig. 10. (A): Calculated mean molar fractions of c- IND (FcIND) or cNAP (FcNAP) ± SD (n = 3) after 56 d and 112 d of storage in dependence of the naproxen molar fraction (NMF) in the samples. (B) Calculated mean residual naproxen molar fractions in the co- amorphous phase (NMFresidual) ± SD (n = 3) after 56 and 112 d of storage in dependence of the naproxen molar fraction in the samples (NMF).

ated and revealed that the eutectic point, reported at an NMF of 0.55 [25] is located in this area. However, the eutectic point was also determined in the present study by evaluation of the heats of fusion of the eutectic peaks of differently composed c-IND/ cNAP physical mixtures. The resulting graphs in Fig. 11 intersect at an NMFP of 0.596 and thus reveal an eutectic point at an NMF of 0.6 rather than 0.55 [25] in the co-amorphous phase. Therefore,

the best physical stability is found exactly at the eutectic composition of co-amorphous NAP/IND. It is reported that if two solids A and B form a eutectic system, the compounds are miscible in the molten state and immiscible in the solid state [32]. In the molten state, it is well known that the interactions between the compounds A and B are greater compared to the interactions between the like components [33,34] and for

A. Beyer et al. / European Journal of Pharmaceutics and Biopharmaceutics 109 (2016) 140–148

H Eutectic peaks (J/g)

200

P = 0.596

180 160

y = 268x – 6 R² = 0.999

140

y = -370x + 374 R² = 0.997

120 100 80 60 40 20 0 0

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

1

NMF Fig. 11. Heats of fusion ± SD (n = 3) of the eutectic peaks in dependence of the naproxen molar fraction (NMF) in the respective cNAP/c-IND physical mixtures.

the eutectic composition it was reported that this interaction is very strong [35]. Therefore, it is expected that the interaction between NAP and IND might also be very strong at the eutectic composition in both the molten state and the resulting solidified co-amorphous state. From the findings reported in this study, it is thus suggested, that the most stable co-amorphous blend is likely to be found at the eutectic point of this mixture. 4. Conclusion In conclusion, the obtained data revealed that intermediate cooling after melt-quenching and a naproxen molar fraction (NMF) in the co-amorphous phase of 0.6 results in the physically most stable co-amorphous NAP/IND samples. Furthermore, coamorphous samples with NMFs different from 0.6 recrystallized predominantly under formation of either crystalline naproxen or c-indomethacin until an NMF of again 0.6 in the residual coamorphous phase was reached. Interestingly, the blend ratio of 0.6:0.4 represents the eutectic composition of the crystalline NAP/c-IND system, which indicates that the eutectic point may play an important role for the stability of binary co-amorphous systems. In conclusion, the physical stability of co-amorphous NAP/IND can be significantly improved if suitable preparation parameters and the optimal phase composition are chosen. These findings might also be applicable to other co-amorphous systems. Acknowledgments The authors would like to thank Isabelle Nevoigt from the Institute of Inorganic and Applied Chemistry at the University of Hamburg for her kind help with the X-ray measurements. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ejpb.2016.10.002. References [1] T. Vasconcelos, B. Sarmento, P. Costa, Solid dispersions as strategy to improve oral bioavailability of poor water soluble drugs, Drug Discovery Today 23–24 (2007) 1068–1075. [2] D. Engers, J. Teng, J. Jimenez-Novoa, P. Gent, S. Hossack, C. Campbell, J. Thomson, I. Ivanisevic, A. Templeton, S. Byrn, A solid-state approach to enable early development compounds: Selection and animal bioavailability studies of an itraconazole amorphous solid dispersion, J. Pharm. Sci. 9 (2010) 3901–3922. [3] Y. Kawabata, K. Wada, M. Nakatani, S. Yamada, S. Onoue, Formulation design for poorly water-soluble drugs based on biopharmaceutics classification system: basic approaches and practical applications, Int. J. Pharm. 1 (2011) 1–10.

147

[4] G.L. Amidon, H. Lennernäs, V.P. Shah, J.R. Crison, A theoretical basis for a biopharmaceutic drug classification: the correlation of in vitro drug product dissolution and in vivo bioavailability, Pharm. Res. 3 (1995) 413–420. [5] B.C. Hancock, G. Zografi, Characteristics and significance of the amorphous state in pharmaceutical systems, J. Pharm. Sci. 1 (1997) 1–12. [6] P. Karmwar, K. Graeser, K.C. Gordon, C.J. Strachan, T. Rades, Investigation of properties and recrystallisation behaviour of amorphous indomethacin samples prepared by different methods, Int. J. Pharm. 1–2 (2011) 94–100. [7] R. Laitinen, K. Löbmann, C.J. Strachan, H. Grohganz, T. Rades, Emerging trends in the stabilization of amorphous drugs, Int. J. Pharm. 1 (2013) 65–79. [8] Abu T.M. Serajuddin, Salt formation to improve drug solubility, Adv. Drug Deliv. Rev. 7 (2007) 603–616. [9] C. Bhugra, R. Shmeis, M.J. Pikal, Role of mechanical stress in crystallization and relaxation behavior of amorphous indomethacin, J. Pharm. Sci. 10 (2008) 4446–4458. [10] P. Karmwar, J. Boetker, K. Graeser, C. Strachan, J. Rantanen, T. Rades, Investigations on the effect of different cooling rates on the stability of amorphous indomethacin, Eur. J. Pharm. Sci. 3 (2011) 341–350. [11] E. Lenz, K. Tarp Jensen, L. Ingerslev Blaabjerg, K. Knop, H. Grohganz, K. Löbmann, T. Rades, P. Kleinebudde, Solid-state properties and dissolution behaviour of tablets containing co-amorphous indomethacin–arginine, Eur. J. Pharm. Biopharm. (2015) 44–52. [12] S. Janssens, G. van den Mooter, Review: physical chemistry of solid dispersions, J. Pharm. Pharmacol. 12 (2009) 1571–1586. [13] W. Zheng, A. Jain, D. Papoutsakis, R.-M. Dannenfelser, R. Panicucci, S. Garad, Selection of oral bioavailability enhancing formulations during drug discovery, Drug Dev. Ind. Pharm. 2 (2012) 235–247. [14] P. Srinarong, H. de Waard, H.W. Frijlink, W.L. Hinrichs, Improved dissolution behavior of lipophilic drugs by solid dispersions: the production process as starting point for formulation considerations, Expert Opin. Drug Deliv. 9 (2011) 1121–1140. [15] A.R. Shrivastava, B. Ursekar, C.J. Kapadia, Design, optimization, preparation and evaluation of dispersion granules of valsartan and formulation into tablets, Curr. Drug Deliv. 1 (2009) 28–37. [16] B.C. Hancock, S.L. Shamblin, G. Zografi, Molecular mobility of amorphous pharmaceutical solids below their glass transition temperatures, Pharm. Res. 6 (1995) 799–806. [17] S. Jayant Dengale, H. Grohganz, T. Löbmann, K. Rades, Recent advances in coamorphous drug formulations, Adv. Drug Delivery Rev. (2016). [18] M. Descamps, J. Willart, E. Dudognon, V. Caron, Transformation of pharmaceutical compounds upon milling and comilling: the role of Tg, J. Pharm. Sci. 5 (2007) 1398–1407. [19] M. Allesø, N. Chieng, S. Rehder, J. Rantanen, T. Rades, J. Aaltonen, Enhanced dissolution rate and synchronized release of drugs in binary systems through formulation: amorphous naproxen–cimetidine mixtures prepared by mechanical activation, J. Control. Release 1 (2009) 45–53. [20] K. Löbmann, H. Grohganz, R. Laitinen, C. Strachan, T. Rades, Amino acids as coamorphous stabilizers for poorly water soluble drugs – Part 1: Preparation, stability and dissolution enhancement, Eur. J. Pharm. 3 (Part B) (2013) 873– 881. [21] K. Löbmann, R. Laitinen, C. Strachan, T. Rades, H. Grohganz, Amino acids as coamorphous stabilizers for poorly water-soluble drugs – Part 2: Molecular interactions, Eur. J. Pharm. 3 (Part B) (2013) 882–888. [22] K. Löbmann, R. Laitinen, H. Grohganz, C. Strachan, T. Rades, K.C. Gordon, A theoretical and spectroscopic study of co-amorphous naproxen and indomethacin, Int. J. Pharm. 1 (2013) 80–87. [23] K. Tarp Jensen, L. Ingerslev Blaabjerg, E. Lenz, A. Bohr, H. Grohganz, P. Kleinebudde, T. Rades, K. Löbmann, Preparation and characterization of spraydried co-amorphous drug–amino acid salts, J. Pharm. Pharmacol. (2015) (n/a). [24] N. Chieng, J. Aaltonen, D. Saville, T. Rades, Physical characterization and stability of amorphous indomethacin and ranitidine hydrochloride binary systems prepared by mechanical activation, Eur. J. Pharm. 1 (2009) 47–54. [25] K. Löbmann, R. Laitinen, H. Grohganz, K.C. Gordon, C. Strachan, T. Rades, Coamorphous drug systems: enhanced physical stability and dissolution rate of indomethacin and naproxen, Mol. Pharm. 5 (2011) 1919–1928. [26] P. Bansal, M. Hall, M.J. Realff, J.H. Lee, A.S. Bommarius, Multivariate statistical analysis of X-ray data from cellulose: a new method to determine degree of crystallinity and predict hydrolysis rates, Bioresour. Technol. 12 (2010) 4461– 4471. [27] F.H. Chung, Quantitative interpretation of X-ray diffraction patterns of mixtures. I. Matrix-flushing method for quantitative multicomponent analysis, J. Appl. Crystallogr. 6 (1974) 519–525. [28] A. Beyer, H. Grohganz, K. Löbmann, T. Rades, C.S. Leopold, Multivariate quantification of the solid state phase composition of co-amorphous naproxen-indomethacin, Molecules 10 (2015) 19571–19587. [29] M. Descamps, E. Dudognon, Crystallization from the amorphous state: nucleation–growth decoupling, polymorphism interplay, and the role of interfaces, J. Pharm. Sci. 9 (2014) 2615–2628. [30] V. Legrand, M. Descamps, C. Alba-Simionesco, Glass-forming meta-toluidine: a thermal and structural analysis of its crystalline polymorphism and devitrification, Thermochim. Acta 1 (1997) 77–83. [31] L.S. Taylor, G. Zografi, Spectroscopic characterization of interactions between PVP and indomethacin in amorphous molecular dispersions, Pharm. Res. 12 (1997) 1691–1698. [32] A.H. Goldberg, M. Gibaldi, J.L. Kanig, Increasing dissolution rates and gastrointestinal absorption of drugs via solid solutions and eutectic mixtures

148

A. Beyer et al. / European Journal of Pharmaceutics and Biopharmaceutics 109 (2016) 140–148

I. Theoretical considerations and discussion of the literature, J. Pharm. Sci. 8 (1965) 1145–1148. [33] M.D. Moore, P.L. Wildfong, Aqueous solubility enhancement through engineering of binary solid composites: pharmaceutical applications, J. Pharma. Innovation 1 (2009) 36–49.

[34] R.P. Rastogi, P.S. Bassi, Mechanism of eutectic crystallization, J. Phys. Chem. 9 (1964) 2398–2406. [35] N. Singh, N.B. Singh, U.S. Rai, O.P. Singh, Structure of eutectic melts; binary organic systems, Thermochim. Acta 1 (1985) 291–293.