Characterization of amorphous celecoxib mixed with plasticizing (TPGS) and anti-plasticizing (PVP) ingredients using Hot Melt Extrusion.

Abstract

Purpose:

Hot melt extrusion (HME) has demonstrated to be an adequate compounding method for poorly-soluble pharmaceutical drugs, as it increases its solubility by fixing its amorphous solid-state using polymers (plasticizing) and other ingredients (non- plasticizing). However, it's amorphous state of the drug and the stability of the amorphous state will greatly depend on its interactions with these (plasticizing or not).

Methods:

In this study, we aimed at characterizing the impact of the combination of plasticizing (TPGS) and anti-plasticizing (PVP) ingredients in amorphous celecoxib prepared using HME in terms of chemical interactions between the components (FTIR, Raman and NMR), viscoelasticity (loss and storage modulus) and required energy for flow (activation energy). Different celecoxib/PVP/TPGS ratios were studied to understand the synergistic effect of PVP and TPGS in inhibiting the crystallization of celecoxib when preparing amorphous dispersions using HME. We aimed at linking the viscoelastic properties of the melt with the resulting amorphous state described by the chemical interactions upon extrusion.

Results:

The amorphous state of celecoxib was evidenced by strengthening of H-bonding between celecoxib and PVP, lack of characteristic crystalline peaks of celecoxib, and deshielding of aromatic protons. The melt was also characterized in terms of viscoelastic temperature dependent behavior (liquid G"; elastic G'), where increasing amounts of TPGS and PVP showed opposites effects; TPGS reduced the viscoelastic response whereas PVP increased it. Calculated melt activation energies (Ea) from the temperature dependent viscosity revealed a threshold of TPGS concentration where samples with 1% w/w of TPGS showed higher flow activation energies (higher Ea) independent of the drug/polymer ratios, compared to samples with higher amounts of TPGS.

Conclusions:

Low drug content combined with anti-plasticizing (PVP) amounts and relatively low plasticizing (TPGS) amounts yields an amorphous dispersion that is characterized with strong H-bonding due to efficient mixing using HME.

INTRODUCTION

Over the past decade, amorphous solid dispersions have received considerable attention due to the increased solubility and bioavailability of poorly soluble drugs [1]. Oral bioavailability is mainly affected by drug solubility, permeability and first pass metabolism. Recent studies found that at least 40% of new drug candidates showed low solubility in water, resulting in low bioavailability because of its crystalline state. Therefore, amorphous solid dispersions is an attractive drug development approach to increase the number of approved drug products in the market [2].

HME is an attractive technology for the production of amorphous dispersions (glassy solutions) with improved aqueous solubility [3]. It may require the use of relatively high temperatures to ensure complete conversion of crystalline API into amorphous form, which is a disadvantage for APIs with poor stability near the melting temperature [4]. As such, polymers and surfactants are used as stabilizing agents within the formulations, facilitating also it’s thermal processing [4, 5].

Polymers are known to inhibit interactions between the drug molecules during HME processing, these drug/drug interactions are inhibited by reducing the molecular mobility of the drug (anti-plasticizing effect) hindering thus crystal growth [3, 5–8]. Chemical shifts in NMR, Raman and FTIR spectrum has provided useful information about API and polymer interactions reflected in chemical shift variations[2, 5, 9]. On the other hand, miscibility of the API with other components specifically surfactants have proven to support crystal growth rate of the drug due to its plasticizing effect[10]. However, ternary mixtures composed of drug, polymer and surfactant result in complex synergistic effect where ultimately the miscibility and concentration will indicate if the surfactant will support or inhibit the crystallization of the drug [10]. Since surfactants and polymers are common excipients used during manufacture of pharmaceutical formulations as amorphous dispersions, it is critical to assess the impact these components may have in the processing and crystal growth rate of the drug when formulated as an amorphous using HME[11].

The purpose of this study is to characterize the impact of the combination of plasticizing (TPGS) and anti-plasticizing (PVP) excipients in the amorphous state of celecoxib prepared using HME. This characterization will be analyzed in terms of chemical interactions between the components (FTIR, Raman and NMR), viscoelasticity (loss and storage modulus) and required energy for flow (activation energy, molecular mobility). We hypothesize that a change in viscosity of the melt due to the presence of the surfactant TPGS will support the PVP mobility between drug molecules. We aim at analyzing the viscosity of the melt in terms of molecular mobility, and thus mixing efficiency during extrusion to hinder the crystalline state of the drug[12]. Recent studies have provided comprehensive information on material processability under conditions analogous to HME processing and demonstrated its potential in the optimization of key HME processing parameters using viscosity and viscoelasticity[4, 13, 14]. The results presented herein focuses on the Arrhenius equation for the determination of the flow activation energy using the measured melt viscosity, and allowing for predictions of the required flow energy for adequate mixing during HME [15–17].

MATERIALS AND METHODS

Materials

Crystalline celecoxib powder was purchased from Attix Pharmaceuticals (T_m = 163 °C)[10]. Polyvinylpyrrolidone K-12 was a generously donated by BASF (T_m = 200 °C). The selected surfactant was D-α-Tocopherol polyethylene glycol 1000 succinate (TPGS) was purchased from Sigma Aldrich. All materials were stored as received in an amber desiccator at ambient temperature. Figure 1 illustrates the molecular structure of celecoxib, PVP and TPGS.

Figure 1.

Molecular structure of (a) Celecoxib, (b) PVP, and (c) TPGS.

Methods

Preparation of physical mixtures

Physical mixtures consisting of Celecoxib, PVP and TPGS were prepared by individually weighing each material and mixing them together. A set of samples with varying concentrations of drug, polymer and surfactant was developed using the Modde software (Umetrics, Sweden), version 11 (Table 1). The concentration range of the different materials was pre-defined. A mixing protocol was developed to mix components. This protocol included side mixing (20 times to the left and 20 times to the right), up and down mixing (10 each), and shaking each for 1 minute. After mixing, the samples were stored in glass vials at ambient temperature in an amber desiccator.

Table 1.

Concentrations of Celecoxib, PVP and TPGS per sample (%w/w).

Sample	Celecoxib (%w/w)	PVP (%w/w)	TPGS (%w/w)
1	30	60	10
2	10	84.5	5.5
3	50	44.5	5.5
4	10	80	10
5	50	49	1
6	30	69	1
7	10	89	1
8	30	64.5	5.5
9	50	40	10
10	30	64.5	5.5

Characterization

XRD

Powder diffraction was performed using a Rigaku X-Ray Powder Diffractometer (XRPD) with a 2-theta (2Θ) range from 5 to 40 with a 2 D/teX detector. All samples were measured at ambient temperature.

FTIR

A Nicolet Continuum FT-IR Spectrometer from Thermo Scientific was used for Infrared spectroscopy in transmission mode. This instrument is equipped with a microscope for selection of sample area to be tested. DTGS KBr detector was used as well as a sample compartment of Right uScope. The FT-IR spectra was measured between 1000-4000 cm⁻¹, however for the purposes of this analysis specific regions where interactions of celecoxib and PVP are known to occur are presented[3]. These are near the carbonyl functional group (C=O), sulfoxide (SO₂) and the amine (NH₂) region.

Raman

A Nicolet Continuum FT-IR Spectrometer from Thermo Scientific was used for Infrared spectroscopy in transmission mode. This instrument is equipped with a microscope for selection of sample area to be tested. DTGS KBr detector was used as well as a sample compartment of Right uScope. The FT-IR spectra was measured between 1000-4000 cm⁻¹, however for the purposes of this analysis specific regions where interactions of celecoxib and PVP are known to occur are presented[3]. These are near the carbonyl functional group (C=O), sulfoxide (SO₂) and the amine (NH₂) region. Changes in bandwidth of the carbonyl band was calculated at half of the band intensity for each sample.

Nuclear Magnetic Resonance (NMR)

¹H NMR spectrum was used to obtain chemical information about the presence of drug-polymer-surfactants interactions that may occur in solution. Chemical displacements were calculated per spectra of hydrogen protons of pure celecoxib and compared with the chemical displacements of the extrudates. Spectra were recorded on a 500 MHz Bruker spectrometer operating at 500 MHz, using a 5mm multinuclear probe with 30-degree pulses of 3.68 microseconds. The measurement was performed at 25 °C in deuterated chloroform (CDCl₃) using tetramethylsilane (TMS) as an internal standard[5].

Rheology

Rheological characterization of the samples was performed using an Anton Paar Modular Compact Rheometer (MCR) series 302 equipped with one set of parallel plates (diameter= 25 mm) with Peltier temperature control. The gap distance was set at 0.800 mm. All oscillatory measurements were performed within the linear viscoelastic region (LVER). Sample loading consisted of adding approximately two grams of each sample into the lower plate at T= 25 °C, afterwards the solid sample was heated above its melting temperature. When the sample was completely melted, the upper plate was lowered (0.8mm) and any sample excess outside of the plates of the instrument was trimmed. Samples were left at the melt temperature for three minutes before any rheological test was performed. A pre-shear was applied to every sample to erase any memory effect due to sample loading (shear rate=10s⁻¹, T=170 °C, 5 minutes). Loss and storage modulus as well as viscosity was measured while applying a temperature ramp from 170 °C to 60 °C. Some samples could not be measured down to 60 °C due to sample solidification, which impeded the measurement to continue. All samples were measured in triplicates and the average of the three data points is reported.

The Arrhenius equation was used to obtain the activation energies of the samples using the measured temperature dependent viscosity [8].

$$\eta =A_e\frac{E_a}{RT}$$ (1)

Where η is the viscosity; A, a constant; E_a, activation energy; R, gas constant; and T, temperature in K. The activation energy (E_a) was calculated by plotting the viscosity as a function of 1/T [1/K] for each sample. In this study the E_a was analyzed in terms of the required energy to initiate the melt flow when extrusion, thus an indirect measurement of the mixing efficiency while extrusion.

Hot Melt Extrusion

A Thermo Scientific HAAKE MiniLabII hot melt extruder equipped with twin co-rotating counter clock screws was used to extrudate the samples. Samples were extruded with a constant temperature of 150 °C and screw speed of 25 min⁻¹. Extruded samples were collected at the exit right after the end dye. Amorphous solid samples were pulverized and stored in glass vials for subsequent analysis.

RESULTS

XRD

Figure 2 shows the X-Ray diffraction patterns of all the extruded samples, and pure crystalline celecoxib. Pure crystalline celecoxib shows its most characteristic diffraction peaks at specific angles of: 5.4 °, 10.8 °, 16.1 °, 21.1°, 22.2° and 27.0°. Being 16.1 o and 21.1 o the two peaks with the strongest intensity. This agrees with reported diffraction patterns of pure crystalline celecoxib[10]. The diffraction pattern of the extruded samples (Figure 2) is significantly different when compared to the diffraction pattern of pure crystalline celecoxib. The characteristic diffraction peaks of pure crystalline celecoxib disappeared when processed by HME, and the appearance of a halo diffraction pattern upon extrusion confirms in general the absence of long range crystalline arrangement within the extruded samples. Sample 4 shows two small intensity peaks at angles (2theta) 19.2 ° and 22.22 °, indicative of some crystalline or polymorphic arrangement of the celecoxib molecules in this sample.

Figure 2.

X-Ray Diffraction of extruded celecoxib samples, PVP, and pure celecoxib.

FTIR

Figure 3 illustrates the Fourier transform Infrared spectra of the sulfoxide (S=O) symmetric and asymmetric stretching region, the carbonyl (C=O) stretching region and the amine (N-H) stretching region of pure crystalline celecoxib and the extruded samples.

Figure 3.

Fourier transform infrared (FT-IR) spectra of the sulfoxide (S=O) symmetric and asymmetric stretching region, the carbonyl (C=O) stretching region and the amine (N-H) stretching region of pure crystalline celecoxib and the extruded samples.

Overall, infrared peak intensity reduction is observed in extruded samples compared to pure crystalline celecoxib. The symmetric and asymmetric S=O stretching intensities having considerable energy differences between the two (asymmetric stretching occurring a higher wavenumbers) are significantly reduced in the extruded samples. The asymmetric S=O stretching was negatively shifted relative to pure crystalline celecoxib evidencing strengthening of hydrogen bonding, except for samples 2 and 7 which could not be determined. Whereas the symmetric S=O stretching shift was drug-concentration dependent, the only samples that were positively shifted were samples with the least amount of celecoxib (2, 4, and 7), the rest of the samples showed no infrared shift relative to pure celecoxib.

Pure crystalline celecoxib revealed two sharp doublets for N-H stretching vibration (3334 and 3228 cm⁻¹) which shifted to higher wavenumbers (3342 and 3253 cm⁻¹) when amorphous (quenching). A broader distribution of H-bonding in amorphous extruded samples was evidenced with the appearance of a broader hump within the N-H stretching FTIR region in all extruded samples, shifted to higher wavelengths (Figure 3)[9]. This positive shift of N-H stretching has been attributed to a decrease of drug/drug interaction due to the presence of PVP, which results in the drug (while amorphous) interacting with PVP by H-bonding through the carbonyl group of PVP[10]. All extruded samples showed the carbonyl band at lower wavelengths (1651cm⁻¹ and 1658cm⁻¹) compared to pure PVP which has a carbonyl stretch at 1681cm⁻¹ [10]. TPGS has been reported to shift even more the carbonyl C=O band, relative to celecoxib/PVP mixtures. Meaning that celecoxib interacts with PVP through H-bonding between the amine (celecoxib) and the carbonyl (PVP) when extruded, this interaction is supported by the plasticizing effect of TPGS. Samples 4 and 7 showed the strongest interaction (lowest C=O wavenumber).

Raman

Raman spectra of pure celecoxib and the extruded samples are illustrated in Figure 4. Only the regions where the sulfur dioxide (SO₂), trimfluormethyl (CF₃) and amine (NH₂) bands are illustrated, these are near 1,160.62 cm⁻¹, 1,229.08 cm⁻¹ and 1,550.17 cm⁻¹ respectively. Overall, extruded formulations showed a significant reduction in the intensity of the SO₂, NH₂ and CF₃ Raman bands compared to pure crystalline celecoxib (Figure 4). Within these bands there is significant band broadening, which combined with the reduction in band intensity evidences a disordered state of celecoxib. This effect is more noticeable in the SO₂ and CF₃ bands, compared to the NH₂ bands. The SO₂ band decreased mostly to a plateau within this region. The CF₃ region having a relatively high baseline made it somewhat difficult to quantify a band reduction, specifically in samples 1 and 6. There was no significant SO₂, CF₃ and NH₂ Raman band position changes in the extruded samples compared to the pure crystalline celecoxib.

Figure 4.

Raman spectra of extruded celecoxib samples, and pure components (celecoxib, PVP, TPGS). (left) Symmetric sulfoxide (SO₂) and trifluoromethyl (CF3), (right) amine functional group (NH3).

Nuclear Magnetic Resonance (NMR)

Figure 5 illustrates the sulfonamide group −SO₂ NH₂ celecoxib protons mostly responsible for the chemical interactions between celecoxib and PVP. Samples with the largest sulfonamide displacement were samples with the highest amount of PVP (Samples 2, 4, and 7), independent of the amount of TPGS. Suggesting that changes in the electron density of the sulfonamide of celecoxib when extruded is mainly attributed to its interaction with PVP, resulting in the shifting of the aromatic protons (H-1b) to higher chemical shifts (deshielding). As the amount of PVP is decreased, drug/polymer interactions also decrease (Figure 4). Thus, relatively high amounts of PVP result in less drug/drug interactions due to drug/polymer interactions. Sample 5 showed the least chemical displacement Δδ = 0.0016 (Figure 4), indicating the least interaction with the polymer.

Figure 5.

NMR¹ H spectrum of the celecoxib sulfonamide group protons of extrudates.

Rheology

Storage (G’) and Loss (G’) Modulus

Figure 6 (left) shows the Storage Modulus of all the extruded samples. All samples were characterized within a temperature range of 170-60 °C. All samples showed an increase of the elastic modulus as the applied temperature was decreased. Meaning that the elastic character of the melt being temperature dependent increases as the applied temperature is decreased; i.e. storage modulus inversely proportional to temperature.

Figure 6.

Storage modulus of physical mixtures of celecoxib, PVP and TPGS as a function of temperature (170-60 °C).

Figure 6 illustrates the plasticizing effect of TPGS on the viscoelastic plot of the samples. When combining the samples keeping the amount of TPGS constant, the storage modulus plot (elastic character) increases as the amount of polymer present increases while cooling from the undercooled melt state. Sample 7 showed the highest storage modulus behavior throughout the measured temperature range in comparison of the rest of the samples. Sample 9 allowed measurements down to 85 °C, this sample having 10% of TPGS had the second lowest elastic behavior near the melt temperature and throughout the temperature range studied.

Figure 6 (right) shows also Loss modulus of all the samples as a function of temperature. The loss modulus follows the similar behavior as the storage modulus in terms of the increase of the moduli as the amount of PVP increases.

Combining the samples by fixing the amount of TPGS results in an increase of the moduli when the amount of PVP increases. Additionally, the plasticizing effect of TPGS is observed as samples with the highest amount of TPGS have the lowest loss moduli (liquid character). As the amount of TPGS increases the loss modulus plot of each sample increases also.

Activation Energy

Table 2 shows the calculated melt flow activation energy of the samples in decreasing order. Sample 5 had the highest activation energy, followed by samples 6 and 7, these are the samples with lowest percentage of surfactant (1%w/w). Followed by samples with 30% w/w celecoxib (1 and 10), 10%w/w celecoxib (4 and 2) and 50%w/w celecoxib (3 and 9). Table 2 illustrates a TPGS concentration apparent threshold, where above 1%w/w (5.5%w/w and 10%w/w) TPGS does not appears to have an impact in the required energy of flow. However, at 1%w/w TPGS (lower plasticizing effect) drug/polymer ratios dictate the required energy of flow of the melt.

Table 2.

Calculated melt flow Activation Energy of each sample using Arrhenius equation (equation 1), samples are arranged by decreasing energy.

Sample	Drug/PVP/TPGS Concentration	Activation Energy (Ea) (KJ/mol)
5	50 CEL /49 PVP/ 1 TPGS	276.85
6	30 CEL /69 PVP/ 1 TPGS	276.22
7	10 CEL /89 PVP/ 1 TPGS	271.81
1	30 CEL /60 PVP/ 10 TPGS	239.40
10	30 CEL /64.5 PVP/ 5.5 TPGS	233.24
4	10 CEL /80 PVP/ 10 TPGS	225.81
2	10 CEL /84.5 PVP/ 5.5 TPGS	224.07
3	50 CEL /44.5 PVP/ 5.5 TPGS	220.24
9	50 CEL /40 PVP/ 10 TPGS	199.187

DISCUSSION

Amorphous dispersions consisting of celecoxib mixed with an anti-plasticizing polymer (PVP) and a plasticizing surfactant (TPGS) were prepared by means of HME. Different celecoxib/PVP/TPGS ratios were studied to understand the synergistic effect of PVP and TPGS in inhibiting the crystallization of celecoxib when preparing amorphous dispersions using HME [10].

The amorphous state of celecoxib was evidenced by strengthening of H-bonding between celecoxib and PVP, lack of characteristic crystalline peaks of celecoxib, and deshielding of aromatic protons [3, 9, 18]. Our results show that the role of TPGS was to mainly facilitate the mobility of the melt to allow optimum H-bonding between celecoxib and PVP, increasing thus the amorphous state of celecoxib in the extrudates [10, 19]. The already reported synergistic effect of PVP and TPGS in decreasing the crystallization rate of celecoxib is suggested to be the result of the increase in the liquid character (loss moduli) of the samples with TPGS content increasing the overall molecular mobility during extrusion [20]. Allowing thus the PVP molecules to inhibit celecoxib/celecoxib interactions, and thus crystallization. However, an apparent threshold of TPGS concentration was observed where high melt activation energies (E_a) are obtained independent of the drug/polymer ratios.

The calculated activation energies (E_a) of the melt provided fundamental understanding of the kinetics of celecoxib when mixed with PVP and TPGS in terms of the required energetic barrier for flow to be initiated. Initiation of flow is important as adequate mixing in HME will greatly depend on the mobility of the molecules past each other due to the mechanical (screws) and thermal energies being applied [8]. Being HME mixing a melt kinetic process, a flow energy barrier must be exceeded for celecoxib to be transferred to the carbonyl group of PVP to thus successfully inhibit its crystallization when cooled down. This is where the TPGS plasticizing effect becomes significant and yet complex. Low concentration of TPGS (1% w/w) and thus low plasticizing effect had the highest energetic flow barrier (high E_a) of all samples independent of drug/PVP ratios. Meaning that these samples require higher mixing energies in order to transfer celecoxib molecules into the adequate carbonyl PVP group and inhibit its crystallization; i.e. E_a and TPGS inversely proportional relationship [21]. Higher TPGS concentrations (5%w/w and 10%w/w) result in more complex behavior with the E a demonstrating to be more dependent on the drug/polymer ratio than with TPGS. Overall, it PVP impact is in increasing the flow activation energy of celecoxib, which may be attributed to the hindering of the celecoxib molecules resulting thus in higher flow energetic barriers[8, 15].

CONCLUSION

We have described an apparent link between celecoxib interactions with PVP and TPGS and its processing using HME. We suspect that drug/drug interactions of celecoxib are hindered due to the strong H-bonding with PVP which is greatly dependent on the plasticizing effect of TPGS. Samples with the lowest amount of the drug resulted to have the strongest H-bonding, compared to the rest. Among these samples, sample 2 had the lowest E a of flow. Pointing towards low drug content combined with high anti-plasticizing (PVP) amounts and relatively low plasticizing (TPGS) amounts will yield an amorphous dispersion that is characterized with strong H-bonding due to efficient mixing using HME.