Hydroxypropyl Methylcellulose-Controlled Crystallization of

Sep 5, 2008 - Citation data is made available by participants in Crossref's Cited-by Linking service. For a more comprehensive list of citations to th...
1 downloads 0 Views 1MB Size
Hydroxypropyl Methylcellulose-Controlled Crystallization of Erythromycin A Dihydrate Crystals with Modified Morphology Sabiruddin Mirza,*,† Inna Miroshnyk,† Jyrki Heina¨ma¨ki,† Jukka Rantanen,‡ Osmo Antikainen,† Pia Vuorela,§ Heikki Vuorela,| and Jouko Yliruusi†

CRYSTAL GROWTH & DESIGN 2008 VOL. 8, NO. 10 3526–3531

DiVision of Pharmaceutical Technology, Faculty of Pharmacy, P.O. Box 56, FIN-00014, UniVersity of Helsinki, Finland, Department of Pharmaceutics and Analytical Chemistry, Faculty of Pharmaceutical Sciences, UniVersity of Copenhagen, Denmark, Department of Biochemistry and Pharmacy, Åbo Akademi UniVersity, Turku, Finland, and DiVision of Pharmaceutical Biology, Faculty of Pharmacy, UniVersity of Helsinki, Finland ReceiVed August 11, 2007; ReVised Manuscript ReceiVed June 30, 2008

ABSTRACT: Crystallization in the presence of pharmaceutically accepted excipients as additives was investigated as a tool for morphological crystal engineering of active pharmaceutical ingredients. Recrystallization of a model drug substance, erythromycin A dihydrate, was carried out by a precipitation technique in the presence of varying amount of hydroxypropyl methylcellulose (HPMC). In contrast to the reference crystals, the crystals grown in the presence of HPMC exhibited regular shape, which varied with the concentration of the additive present, thus indicating that HPMC is an effective habit modifier for erythromycin A dihydrate. On the basis of the crystal surface chemistry, the mechanism responsible for the change in morphology is proposed. Finally, the effect of crystal habit modification on compaction behavior of erythromycin A dihydrate is demonstrated. Introduction Active pharmaceutical ingredients (APIs) are most often administered in crystalline forms that are more stable and easier to handle than amorphous materials. Accordingly, a key stage for manufacturing APIs is crystallization, which defines physical properties of a drug (crystal structure and degree of crystal imperfection), as well as crystal shape and size. Crystal shape, often referred to as morphology or habit, is one of the most important descriptors of pharmaceutical solids since it influences powder properties of significant technological and biopharmaceutical importance.1,2 For example, crystal habit influences powder flow (equidimensional crystals flow better than needles or plates), powder blending and mixing (similar particle sizes and shapes are preferable), and compactability in tablet manufacture (particle orientation and cohesion, capping, and lamination of the tablets).3 The control over crystal habit is therefore a critical issue in the development of the optimal delivery system for a particular drug. In general, crystal habit modification can be induced by varying any of external crystallization parameters such as solvent, supersaturation, temperature, and additives.4-6 For pharmaceuticals, the solvent-induced crystal habit modification has been traditionally employed.7-10 However, this approach is limited by the solvent toxicity and cost, crystallization efficiency, and the purity requirements of the final product.11 In this context, the use of additives (e.g., surface-active agents12 and polymeric materials11,13) as crystal habit modifiers of APIs can be a beneficial alternative. Moreover, employing pharmaceutically accepted excipients as additives would be the most practical option for such a highly regulated industrial sector as the pharmaceutical industry. In this study, crystallization in the presence of pharmaceutically accepted excipients as additives is investigated as a tool * To whom correspondence should be addressed. E-mail: [email protected]. Phone: (358) 919159151. Fax: (358) 919159144. † Division of Pharmaceutical Technology, University of Helsinki. ‡ University of Copenhagen. § Åbo Akademi University. | Division of Pharmaceutical Biology, University of Helsinki.

Figure 1. Molecular structures for the model drug, erythromycin A (A), and the polymeric additive, HPMC (B). The relevant oxygen atoms are numbered in panel A.

for morphological crystal engineering of APIs. Erythromycin A (Figure 1A), a widely prescribed macrolide antibiotic, was selected as a model compound. It crystallizes from aqueous solutions as a dihydrate (EM · DH) in the orthorhombic space group P212121 with unit cell parameters a ) 9.183 Å, b ) 9.632 Å, and c ) 47.151 Å.14 EM · DH crystals grown from aqueous solutions exhibit a plate-like morphology.15 Crystals with this kind of habit demonstrate poor handling and processing characteristics including flowability and compactibility. Poor compaction behavior of APIs is considered to be a major source of batch failures in the pharmaceutical industry. Hence, more equidimensional crystals are usually preferred for tablet manufacturing as they have better compaction performance. The compaction behavior of drugs to be administered in high doses, such as EM · DH, is of special interest since the most desirable technique to produce tablets loaded with these APIs is direct compression. This tableting procedure requires good powder flow properties, uniform mixing between the drug and the excipients, and the ability to consolidate and bond under pressure and maintain interparticle bonds on ejection from the tablet

10.1021/cg7007599 CCC: $40.75  2008 American Chemical Society Published on Web 09/05/2008

Crystallization of Erythromycin A Dihydrate Crystals

Crystal Growth & Design, Vol. 8, No. 10, 2008 3527

Table 1. Compositions of the Crystallization Medium Studied

sample code

concentration of HPMC solution added (%)

A B C

0.01 0.05 0.1

concentration of HPMC in the crystallization medium (g/L)

(wt %)

0.1 0.5 1

0.45 2.25 4.5

press.3 All these properties are directly related to crystal habit and/or particle size distribution. Thus, there is a significant need to develop morphological crystal engineering protocols to enhance processing behavior of pharmaceuticals. The preliminary additive screening experiments in our laboratory have showed that hydroxypropyl methylcellulose (HPMC; Figure 1B), a commonly used pharmaceutical excipient, can be a potential habit modifier for EM · DH. The primary goal of this study was therefore to evaluate the effect of HPMC on the crystal habit of EM · DH during crystallization from solution. Compaction tests with the modified crystals were further performed to estimate whether or not the crystal habit modification can affect tableting behavior of EM · DH. Experimental Section Materials. Erythromycin A dihydrate (EM · DH, MW ) 770 g/mol) (Sandoz, Italy) was used as received. Hydroxypropyl methylcellulose (HPMC; Methocel E 4000, MW ∼ 86 kDa, Dow Chemical Company, U.S.A.) was used as additive. Ethanol of analytical grade and purified water (Ph. Eur.) were used as solvents. Crystallization Procedure. Batch crystallizations (n ) 3) were carried out in 500 mL glass beakers at ambient temperature. The crystals were precipitated by slow pouring a saturated solution of the drug in ethanol into an aqueous solution of the additive of varying concentrations under constant stirring. The final additive concentrations in the crystallization medium studied were 0 (thereafter this sample will be referred to as reference), 0.1, 0.5, or 1 g/L (Table 1). To prevent solvate formation15 the experiments were completed at a solvent/antisolvent ratio of 1:9 (v/v). The solids precipitated were vacuum filtered, dried at 25 °C for 24 h, packed in sealed glass vials, and stored at ambient temperature and RH. Solid-state analysis was performed within 48 h of the sample collection. The drug content after recrystallization was assayed by HPLC according to the method described in the European Pharmacopoeia (5th ed.). Powder X-ray Diffractometry (PXRD). X-ray diffraction studies were done using a theta-theta diffractometer (D8 Advance, Bruker axs GmbH) fitted with a scintillation counter. The experiments were performed in symmetrical reflection mode using a Cu KR radiation source (wavelength 1.54 Å) with Go¨bel Mirror bent gradient multilayer optics. Data points were collected between 5 and 40° 2θ with steps of 0.1° and a measuring time of 1 s/step. Differential Scanning Calorimetry (DSC). A DuPont differential scanning calorimeter (Model 910S, TA Instruments Inc.) equipped with a data station (Thermal analyst 2000, TA Instruments Inc.) was used to determine the DSC curves representing the rates of heat flow with respect to temperature. The temperature was calibrated with benzophenone and indium (melting points at 48.0 and 156.6 °C, respectively). Enthalpy calibration of the DSC signal was performed with indium (10 mg, 99.999% pure, and heat of fusion 28.4 J g-1). Samples were analyzed in open aluminum pans at a heating rate of 10 °C min-1 under static air. Thermogravimetric Analysis (TGA). The loss of mass from the samples was monitored with a Mettler Toledo TA8000 system equipped with a TGA850 thermobalance. Samples (∼10 mg) were analyzed in an open aluminum pan under nitrogen purge (50 mL min-1). A heating rate of 10 °C min-1 was employed, and the temperature range analyzed was from 25 to 120 °C. Scanning Electron Microscopy (SEM). SEM images were recorded with a Zeiss DSM-962 (Carl Zeiss, Oberkochen, Germany) scanning electron microscope at an acceleration voltage of 4-15 kV and were used to examine crystal morphology. Prior to the capturing, the samples were attached to double-sided carbon tape and coated with 20 nm platinum using an Agar sputter coater B7304 (Agar Scientific Ltd., Stanstedt, U.K.).

Figure 2. SEM images of erythromycin A dihydrate crystals grown in the presence of various concentration of HPMC: (A) 0 (reference), (B) 2.25 (B crystals), and (C) 4.5 wt % (C crystals). Molecular Modeling. The single crystal X-ray diffraction data for EM · DH14 (refcode: NAVTAF) were retrieved from the Cambridge Structural Database (CSD). The theoretical powder X-ray diffraction pattern was calculated, and the Miller index of each diffraction peak was assigned using the Diffraction-Crystal Module of Cerius.2 The theoretical growth form was computed by using molecular simulation software (Mercury, Cambridge Crystallographic Data Center, CCDC, Cambridge, U.K., v. 1.5) according to the Bravais-Friedel-DonnayHarker (BFDH) theory. The same software was used for crystal structure visualization. Compaction Tests. Crystal compaction behavior was evaluated with an instrumented eccentric tablet machine (Korsch EK0, Erweka Apparatebau, Germany). For each experiment, the powder (250 ( 10 mg) was manually poured into a die and compacted at various compression pressures using 9 mm diameter flat-faced punches. Thereafter, the crushing strength of the tablets was measured using a tablet hardness tester (Erweka, Germany). The results are represented as the mean and standard deviations of three determinations.

Results and Discussion Morphology of the Recrystallized Erythromycin A Dihydrate Crystals. The SEM images of EM · DH recrystallized in the absence and in the presence of the additive (Figure 2) clearly show that the crystal morphology changes as a function of HPMC concentration in the crystallization medium. The reference crystals (Figure 2A) exhibited a range of shapes, from irregular to acicular and plate-like. A possible explanation for such diverse morphology is fast crystal growth (probably because of the high degree of supersaturation) preventing the formation of fully developed crystals. The additive-treated crystals (Figure 2B,C), in contrast, exhibited characteristic morphology, with the exception of the A crystals (not shown) precipitated at relatively low concentration (0.45 wt %) of HPMC present. Likewise to the reference sample, the shape of the latter crystals was rather irregular. This indicates that, under the present experimental conditions, the lowest HPMC concentration studied was unable to affect the crystal habit of EM · DH. Upon further increase in the polymer concentration to 2.25 and 4.5 wt % the plate-like (Figure 2B) and elongated plate-like (Figure 2C) crystals were produced, respectively. This result may indicate that HPMC affects the growth rates of the EM · DH crystal faces differently.

3528 Crystal Growth & Design, Vol. 8, No. 10, 2008

Mirza et al.

Figure 3. (A) Erythromycin A dihydrate (EM · DH) crystal shape predicted by BFDH model along with the unit cell and (B) the observed morphology of the crystals grown in the presence of 2.25 wt % HPMC (B crystals).

Morphological Model of Erythromycin A Dihydrate Crystal. Particular morphological effects are largely structurally driven and for this reason are specific to each crystallizing system. To better understand these effects with respect to the present system, the predicted BFDH morphology of the EM · DH crystal was visualized (Figure 3A). Application of the BFDH theory represents a quick approach for identifying the crystallographic forms {hkl} most likely to constitute a crystal habit. The relative growth rate of a face, based on the BFDH model, is inversely proportional to the interplanar spacing d, and hence, the most morphologically important faces of the crystal are those that have the biggest d-values. For EM · DH, these planes are (002) (d ) 23.6 Å), (011) (d ) 9.4 Å), and (101) (d ) 9.0 Å), as determined by indexing the single crystal data. Accordingly, the predicted shape of EM · DH is plate-like, slightly elongated along b direction, and bounded by the faces (002), (011), and (101). In addition, small (110) faces emerge as the boundary between the (011) and (101) crystal faces. The comparison of this model with the observed morphology of the reference crystals (i.e., grown in the absence of the additive) was complicated because of the rather irregular shape of the latter crystals. In contrast, it was striking to observe that the morphology of the crystals grown in the presence of 2.25 wt % HPMC (Figure 3B) was in reasonable agreement with the BFDH model. Generally, the observed habit of a crystal is a consequence of the relative growth rates of its faces. The latter result therefore suggests that under these specific experimental conditions the relative growth rate of the individual crystal faces was not changed by the additive present, although the overall crystal growth velocity was apparently suppressed. Solid-State Characterization of the HPMC-Modified Crystals. It is well-known that the changes in crystal habit may be indicative of a phase transition. Hence, to investigate the effect of HPMC present in the crystallization medium on the solid-state and crystallinity of EM · DH, the samples were analyzed by means of powder X-ray diffraction (PXRD), differential scanning calorimetry (DSC), and thermogravimetric analysis (TGA). Comparison of the computed PXRD pattern for EM · DH and the experimental patterns (Figure 4) shows that apart from intensity differences, which can be attributed to preferred orientation in the samples, there is an excellent correspondence between the peak angular positions. Therefore, it can be concluded that no phase transition occurred during crystallization and all crystals represent phase-pure EM · DH. Further, the reflections associated with the dominant crystal faces were examined more closely. The reflections corresponding to the (002) crystal plane appear at around 3.7° 2θ and hence

Figure 4. Representative PXRD pattern of the HPMC-modified crystals of erythromycin A dihydrate (EM · DH) vs commercial sample and reference. Bars represent a simulated PXRD pattern (CSD refcode: NAVTAF).

could not be detected with wide angle PXRD. The diffraction peak associated with the (011) plane is poorly resolved because of its low relative intensity. The intensity of the diffraction peak corresponding to the (101) crystal plane in the patterns of the additive-treated crystals was observed to slightly increase as compared with that of the commercial sample or the reference. This can be ascribed to partial dehydration of EM · DH, as has been previously demonstrated.16 Furthermore, the analysis of TGA results revealed a trend toward slightly decreased (by ∼0.6%) water content for the HPMC-treated crystals (Table 2). Overall, this experimental data suggests that HPMC substitutes terminated water molecules in the EM · DH crystal faces since the presence of the polymer was the only difference between the crystallization protocols of the crystals compared. Finally, possible interactions between the drug and polymer were examined by DSC. The analysis of the DSC traces of the samples shows that some melting point depression (by ∼6 °C) and reduction in enthalpies was a common tendency for the HPMC-modified crystals (Table 2). These reductions in enthalpy of melting may be attributed to the presence of amorphous regions in the crystals or to the weakening and disruption of the crystal lattice and order.1 However, no crystalline disorder was detected by PXRD, indicating that the changes take place mainly at the particle surface. In addition, there was a good correlation between the true density of the single crystal (1.23 g/cm3) and the HPMC-modified crystals (∼1.2 g/cm3), suggesting a high degree of crystallinity of these samples. Therefore,

Crystallization of Erythromycin A Dihydrate Crystals

Crystal Growth & Design, Vol. 8, No. 10, 2008 3529

Table 2. Thermal Characteristics of the Unmodified and HPMC-Modified Crystals of Erythromycin A Dihydrate (EM · DH) sample

onset (and peak) temperatures of dehydration (°C)

enthalpy of dehydration (J g-1)

onset (and peak) temperatures of melting (°C)

enthalpy of melting (J g-1)

TGA weight loss (%)a

commercial EM · DH reference crystals A crystals B crystals C

65.2 ( 1.6 (79.5 ( 0.7) 66.8 ( 1.9 (77.6 ( 0.9) 64.1 ( 1.2 (78.1 ( 2.0) 66.2 ( 0.9 (77.2 ( 2.4) 67.4 ( 1.0 (77.4 ( 2.3)

118.8 ( 0.6 113.4 ( 1.0 106.2 ( 1.6 106.9 ( 3.1 108.4 ( 2.7

127.3 ( 1.4 (134.8 ( 1.6) 127.6 ( 1.2 (133.3 ( 1.5) 122.5 ( 1.9 (129.7 ( 2.8) 122.0 ( 1.3 (129.2 ( 2.1) 121.6 ( 0.8 (130.8 ( 1.4)

5.2 ( 0.07 4.0 ( 0.08 3.3 ( 0.10 2.7 ( 0.07 2.9 ( 0.09

4.5 ( 0.14 4.2 ( 0.10 3.8 ( 0.15 3.9 ( 0.06 4.0 ( 0.11

a

The theoretical water content of EM · DH is 4.7% (w/w). Table 3. Surface Chemistry of the Dominant Crystal Faces (hkl) of Erythromycin A Dihydrate (EM · DH)a number of functional groups exposed to crystal faces (hkl) (002a) (002b) (011) (101)

OH

C2H5

CH3

COC

OCH3

N(CH3)2

H2O

2 2 2 1

1 1

1 4 5

3 1

1 1

1 2 1

a

Two possible terminations, (002a) and (002b), of the largest (002) crystal face are presented.

Figure 5. Crystal lattice of erythromycin A dihydrate (EM · DH) viewed down the a-axis. The dotted lines show the hydrogen-bonding network; for clarity, the hydrogens are omitted.

the peak shifts and enthalpy deviations revealed by DSC were attributed to interactions between the EM · DH crystal surface and the adsorbed additive. Quantification of polymers, including HPMC, is a challenging task because of the lack of UVadsorbing chromophores and often requires the development of alternative detection methods,17 which is out of the scope of the present study. Hence, the amounts of the additive (AA, %) possibly adsorbed by the crystal surfaces were quantitatively estimated by using the results of HPLC determinations of erythromycin A:

AA ) 100 - Drug Content(%)

(1)

On the basis of these calculations, it was approximated that small amounts of the polymer (∼0.2-0.3 wt %) may associate with the EM · DH crystal faces when the crystallization is performed in the presence of 2.25 or 4.5 wt % of HPMC. Possible Mode of the Association of HPMC with Erythromycin A Dihydrate Crystals. Considering the general effect of the additives on crystallization kinetics, it can be assumed that with addition of HPMC to the crystallization medium the mean growth rate is suppressed and altered growth rates of individual crystal faces results in crystal habit modification. Theoretically, EM · DH is expected to grow as plates with the fastest growth along the b-axis as this direction has the strongest drug-drug intermolecular interactions18 and the shortest unit cell dimension (Figure 5). The growth rate along the a-axis should be of the same magnitude since the intermolecular interactions along this direction are almost equally strong18 and the values of the a- and b-parameters of the unit cell are approximately the same. (Comprehensive quantitative analysis of EM · DH crystal packing will be addressed in a subsequent publication.) The morphology of the EM · DH crystals grown in the presence of 2.25 wt % of the additive resembles the BFDH model. Upon increase in the additive concentration, EM · DH grows as more elongated crystals, apparently with decreased growth rate along the b-axis. This change in morphology suggests specific interactions between the polymer and the functional groups on the {011} faces of the crystal.

The well-defined arrangement of molecules on crystal surfaces offers a means to probe molecular recognition events taking place at interfaces during crystallization. Because of the anisotropy of the EM · DH crystal structure (see Figure 5), the terminating surfaces of the crystal may differ substantially. The surface chemistry analysis of the dominant faces of EM · DH supports this statement (Table 3 and Figure 6). In particular, the largest (002) face of the crystal was found to have two possible terminations, (002a) and (002b) (Figure 6A). In the latter case, the cleavage plane would dissect the H-bond O6-H · · · O7-H between the drug molecules as well as the H-bond O6-H · · · O15-H between the drug and water molecules, which seems rather unlikely for energetic reasons. Hence, this face is anticipated to be -N(CH3)2 terminated and develops the most hydrophobic and less energetic (002a) crystal face. The (011) face is the most hydrophilic face of the EM · DH crystal, with the two water molecules being exposed to this face. HPMC has been previously shown to be capable of interacting with the drug molecules via H-bonds.19-21 We assume that the polymer reversibly interacts with the EM · DH crystal surfaces, especially with (011) and (101), during crystallization thus suppressing overall crystal growth. Possible interaction sites of HPMC with the crystal surfaces are those where the water molecules, W1 and W2, are bonded to the drug, as suggested by solid-state analysis of the samples. Specifically, the hydroxyl groups of HPMC may interact with the EM · DH surface via the H-bonds O6 · · · HO-HPMC, O12 · · · HO-HPMC, or O4H · · · O-HPMC (Figure 6D). At relatively low concentrations of the additive, its suppressing effect on the crystal growth is hindering because of insufficient number of polymer molecules. With increased additive concentration, more polymer molecules are available, and the probability of the drug-polymer interactions is higher. Under these conditions, the preferred adsorption of HPMC at the {011} crystal faces leads to retarded crystal growth along the b and c directions, and this effect is macroscopically evident as elongated EM · DH crystals. Compaction Behavior of the HPMC-Modified Crystals. Mechanical properties of the powdered material can roughly be assessed in terms of compact strength as a function of applied compaction stress.22 To estimate the consequences of the crystal habit modification on the compactibility of EM · DH, powder compacts were prepared and evaluated by plotting their crushing strength versus compaction pressure (Figure 7). The commercial

3530 Crystal Growth & Design, Vol. 8, No. 10, 2008

Mirza et al.

Figure 6. Structure of the dominant crystal faces (hkl) of erythromycin A dihydrate (EM · DH): (A) (002), with two possible terminations, (002a) and (002b), of this face being shown, (B) (101), and (C) (011) and (D) interaction sites of two water molecules, W1 and W2, with erythromycin A depicted on the enlarged image of the (011) face. The hydrogens are omitted for clarity.

Figure 7. Compaction profiles of the HPMC-modified erythromycin A dihydrate (EM · DH) crystals.

sample, reference, and A crystals all produced weak compacts with high tendency to cap, regardless of the compaction pressure applied. For the C crystals, a linear relationship between the crushing strength of the compacts and the compaction force was observed at lower compaction forces (