Trospium Chloride: Unusual Example of Polymorphism Based on

Article pubs.acs.org/crystal

Trospium Chloride: Unusual Example of Polymorphism Based on Structure Disorder Eliška Skořepová,†,* Jan Č ejka,† Michal Hušaḱ ,† Václav Eigner,†,§ Jan Rohlíček,†,§ Antonín Šturc,‡ and Bohumil Kratochvíl† †

Department of Solid State Chemistry, Institute of Chemical Technology Prague, Technicka 5, Prague 6, Czech Republic InterpharmaPraha, a.s. Komoranska 955, Prague 12, Czech Republic § Institute of Physics AS CR, v.v.i., Na Slovance 2, 182 21 Prague 8, Czech Republic ‡

S Supporting Information *

ABSTRACT: The characterization of the solid state properties of an active pharmaceutical ingredient is a necessary step in drug development. One such API is trospium chloride, an anticholinergic drug used to treat incontinence and overactive bladder syndrome. One crystal form is known but, from the crystallographic point of view, has only been described by the X-ray powder diffraction pattern. Because, to the best of our knowledge, no crystal structures of trospium chloride have been published, we decided to prepare and describe different crystal forms. Using single-crystal X-ray diffraction, the structures of two new polymorphs and of one unstable solvate were solved. All forms were obtained from ethanol mother liquor. Both polymorphs exhibited a significant pseudosymmetrical disorder. The polymorphs were highly similar but were distinguished by their symmetry and the resulting single-crystal diffraction behavior. A detailed inspection of the molecular packing revealed that variations in the short-range disorder most likely gave rise to these two sibling polymorphs, which are hard to distinguish by X-ray powder diffraction.

1. INTRODUCTION The characterization of solid forms is of great importance in the pharmaceutical industry.1 Polymorphs, hydrates, solvates, salts, and cocrystals of an API (active pharmaceutical ingredient) have generally different physicochemical properties, such as solubility, dissolution rate, stability, or hygroscopicity. Problems caused by an inadequate understanding of the solid state can lead to an unexpected phase transition and result in an altered pharmacokinetic profile, altered bioavailability, reduced shelf life, or problems in the production of the final drug product.2 Polymorphism is the ability of a substance to exist in various crystal arrangements while the chemical composition of the solid stays unchanged.3 There are two basic categories: packing polymorphism and conformational polymorphism. Packing polymorphs adopt different packing patterns, while conformational polymorphs consist of molecules in different conformations. Of course, it is common that these two properties vary at the same time. From the macroscopic point of view, polymorphs differ in many properties, of which some are easily measured (e.g., melting point, X-ray powder diffraction pattern). However, it is not always as easy to distinguish polymorphs as it would seem. One of the major problems © 2013 American Chemical Society

complicating unambiguous identification of polymorphs is molecular disorder. Molecular disorder is a form of a crystal defect. In an ideal crystal, all unit cells exhibit the same arrangement. However, in real crystals, sometimes a portion of the unit cells is different from the majority of the crystal. In pharmaceutical compounds, orientational and conformational disorders are the most common defects.4 These two types differ in molecular orientation. Orientational disorder causes whole molecules to be oriented differently (e.g., rotation of a solvent in a crystal lattice), while conformational disorder affects only parts of the molecules (e.g., two conformations of a side chain). Of course, different types of disorder can be involved simultaneously. Like polymorphism, disorder can cause similar problems in drug production. Disordered materials with varying ratios of disordered components can have different physicochemical properties5 leading to quality control issues. Received: May 14, 2013 Revised: October 14, 2013 Published: October 17, 2013 5193

dx.doi.org/10.1021/cg4007394 | Cryst. Growth Des. 2013, 13, 5193−5203

Crystal Growth & Design

Article

Republic). To simulate the industrial production of TCl, different combinations of the following growth conditions were used to prepare the samples: • Solvent (mixtures of water and ethanol): φ(EtOH) = 1; 0.99; 0.95; 0.8; 0.5 • Crystallization temperature: −30 °C, 0 °C, 20 °C, 45 °C • Initial concentration of the solution: saturated solution, 30% and 50% surplus of solvent An extensive set of solution-based crystallization samples of TCl was prepared. TCl is industrially crystallized from ethanol, so we varied only water content, solution saturation, and temperature. Samples were analyzed by 13C CP/MAS NMR, DSC, FTIR, and XRPD. Results of these experiments were discussed in detail in a published article19 (see the Introduction). The first three methods detected slight differences (e.g., ±2 °C in melting point), but there were virtually no differences in XRPD. The differences between the samples were explained as a result of different extents of molecular disorder. From some of those experiments, single crystals were obtained. To prepare single-crystals as colorless prisms, the material was dissolved at 70 °C and the solution was allowed to cool spontaneously and crystallize in an open vial at room temperature for 48 h. The growth conditions of the crystals measured by SXRD were as follows: • Form I: EtOH 0.95, RT, saturated solution • Form II: EtOH 0.5, RT, saturated solution • Form III: EtOH 1, −2 °C, 50% surplus of solvent Note: In the search for novel solid forms of trospium chloride when crystallized from ethanol−water mixtures, many crystals were tested (unit cell determination by SXRD). Almost always, Form I was detected. Forms II and III were rare. Form II was observed only in three samples. Discovery of Form III was truly serendipitous, because, out of more than one hundred ethanol based crystallization experiments, ethanol solvate was obtained once, and it was in a mixture with Form I. Unfortunately, despite our great effort, crystallization conditions to reproducibly prepare Forms II and III were not found. Slurry experiments to find the stable polymorph were carried out as follows. Trospium chloride (Form I, 100 mg) was mixed with nbutanol and with methyl ethyl ketone (0.5 mL). Those two solvents were selected because TCl is only mediocrely soluble in them, and it is therefore simple to prepare easy-to-handle slurry experiments. Furthermore, TCl does not form solvates with these solvents. Samples were slurried at 25 °C for 6 weeks. Then the samples were filtered and measured by XRPD. Samples obtained after filtering were fine powders, so further milling was unnecessary. 2.2. Synchrotron Based Powder Diffraction. The measurement was done at the European Synchrotron Radiation Facility, beamline ID31. A sample in a glass capillary was measured in the range of 1° to 35° with a step size of 0.002°. The wavelength was set to 0.79986 Å. 2.3. Single-Crystal X-ray Diffraction. A suitable single-crystal was directly mounted on the goniometer. Data was collected on Xcalibur PX, Cu Kα using ω scans. Data collection, cell refinement, and data reduction were done by CrysAlisPro CCD.24 For structure solution, SIR9225 and Superflip26 were used. The programs CRYSTALS,27 Jana2006,28 and Platon29 were used to refine the structure and analyze the absolute configuration. Twin laws were searched for by ROTAX.30 Molecular graphics were prepared in Mercury31 and Discovery Studio.32 The Form I and Form II structures were solved by direct methods. For both forms, the independent part of the unit cell was built of two heavily overlapping disordered molecules. The investigation of both partially occupied molecules revealed a noncrystallographic symmetry operation, a mirror plane perpendicular to the b axis. Due to the overlaps of many ADPs, the stability of the least-squares refinement was very low. All attempts for refinement as a twin or a virtual supergroup failed. Hence, only the major part of the molecule was refined: the minor part was refined “riding” as an exact mirror copy of the major one; that is, the negative shift was applied to the y coordinate, U12 and U23. The occupancy of the major part was refined

A question arises: Where is the borderline between polymorphism and a sample specific disorder? Recently, multiple articles have been published concerning this issue.4,6−16 The cases of eniluracil11 and promethazine hydrochloride17 illustrate how fuzzy the borderline really is. The first is a story of polymorphs later shown to be disordered variations of the same structure. Conversely, the second compound has two polymorphs, but they primarily differ only at the disorder level. The crystal structure of eniluracil was initially determined from powder diffraction, and two polymorphs were identified. However, Copley et al.11 were able to prepare four singlecrystals under different conditions. They showed that the structural variations observed for the crystals of eniluracil were the result of variable degrees of disorder rather than of polymorphism. In the case of disorder vs polymorphism, disorder wins. For promethazine hydrochloride, the structures of two polymorphs were determined using single-crystal X-ray diffraction. The structures exhibited high similarity in terms of both molecular conformation and crystal packing. A slight variation of cell parameters was attributed to different disorder levels in the aliphatic part of the molecule. This small displacement of the molecules in the crystal lattice was sufficient to induce slight energy, density, and melting point differences between the forms. The authors also suggested that the observed disorder was caused by the intergrowth of polymorphic domains.17In the case of disorder vs polymorphism, polymorphism wins. Trospium chloride (TCl), an anticholinergic used to treat incontinence and an overactive bladder,18 is an example of an API exhibiting a strong influence of disorder-causing problems with polymorphic identification. We recently published a multidisciplinary solid state study of TCl.19 Mainly through solid state NMR, a broad variety of structural phases arising from molecular disorder were observed. With the help of factor analysis, relationships were found between the extent of molecular disorder and the crystallization conditions. Although TCl has been known and manufactured for a long time20 and there are sources mentioning crystal phases of TCl,21−23 until now, no one has been able to prepare single-crystals of sufficient quality for structure determination. In this work, we present the structures of two polymorphs and one solvate of TCl (Figure 1) determined using single-

Figure 1. Molecular formula of trospium chloride (TCl).

crystal X-ray diffraction. In addition, we propose possible distributions of the disordered molecules that explain the differences in unit cells and space groups of both observed polymorphs.

2. EXPERIMENTAL SECTION 2.1. Sample Preparation. Samples were prepared by recrystallization of powdered trospium chloride (Interpharma, Prague, Czech 5194



Article

Figure 2. Rietveld fit of Form I on powder synchrotron data: red line, measured; blue, calculated. as one least-squares parameter; the sum of both occupancy values was fixed to the value of 1. Such an approach led to a significant decrease of R in both structures. The positional and anisotropic thermal parameters of all non-hydrogen atoms were refined. The H atoms were all located in a difference map, but those attached to carbon atoms were repositioned geometrically. The H atoms were initially refined with soft restraints on the bond lengths and angles to regularize

their geometry (C−H in the range 0.93−0.98 Å, O−H to value of 0.82 Å) and Uiso(H) (in the range 1.2−1.5 times Ueq of the parent atom). The positions of hydrogen atoms were refined with riding constraints. No solvent-accessible void was found in any of the structures. 2.4. X-ray Powder Diffraction. Sample was ground and placed into the flat holder. A Philips X’PERT PRO MPD PANalytical diffractometer (Almelo, The Netherlands) with Cu Ka radiation 5195



Article

Table 1. Crystallographic Data form formula formula wt color cryst morphology cryst size (mm3) temperature (K) radiation wavelength (Å) cryst system space group a (Å) b (Å) c (Å) α (deg) β (deg) γ (deg) volume (Å3) Z density (g/mL) μ (1/mm) F(000) θ(min, max) no. unique reflns no. obs. reflns no. of params hmin,max kmin,max lmin,max R_all, wR2_all R_obs, wR2_obs Δρmin, Δρmax (e Å−3) GOF CCDC number

polymorph 1

polymorph 2

ethanol solvate

I C25H30Cl1N1O3 428.98 colorless block 0.23 × 0.30 × 0.43 150 Cu Kα 1.54184 monoclinic P21 9.1551(10) 10.9184(13) 11.0126(12) 90 101.042(11) 90 1080.4(2) 2 1.319 1.776 456.0 4.09, 78.65 4446 2272 273 −11, 11 −13, 13 −13, 13 0.090, 0.160 0.054, 0.105 −0.29, 0.43 0.9983 963575

II C25H30Cl1N1O3 428.98 colorless block 0.20 × 0.38 × 0.49 150 Cu Kα 1.54184 monoclinic P21/c 10.9483(3) 10.9241(3) 18.0149(5) 90 100.685(3) 90 2117.23(10) 4 1.343 1.812 912.0 4.11, 77.71 4459 3253 273 −13, 13 −13, 13 −22, 21 0.068, 0.101 0.054, 0.093 −0.71, 0.70 0.9999 963576

III C27H36Cl1N1O4 474.02 colorless rod 0.27 × 0.51 × 0.59 150 Cu Kα 1.54184 triclinic P1̅ 9.7087(6) 14.3561(8) 18.8051(11) 70.573(5) 88.567(5) 88.179(5) 2470.3(3) 4 1.274 1.632 1016 4.56, 77.87 20122 14010 608 −12, 11 −18, 18 −23, 23 0.075, 0.104 0.048, 0.094 −0.17, 0.31 1.02 963577

(wavelength 1.54180 Å) was used at 40 kV and 30 mA. The samples were scanned at 4−40° 2Θ with a step size of 0.017° 2Θ and a step time of 21.32 s. For samples where a higher resolution was needed, measurement was done on a PANalytical Empyrean powder diffractometer from 4° to 60° 2Θ with Cu Kα 1 radiation monochromatized by the incident beam Johansson monochromator Ge (111). The step size was set to 0.013° 2Θ. The results were analyzed using X’PERT HighScore Plus software.33 2.5. Temperature-Resolved X-ray Powder Diffraction. Samples were ground and placed into the 0.3 mm borosilicate glass capillary. The measurement was done in transmission mode on the PANalytical Empyrean powder diffractometer from 4° to 40° 2Θ with Cu Kα1,2 radiation (λ = 1.54184 Å, focusing mirror, step size was 0.013° 2Θ; irradiated length of the capillary was 20 mm). The device was equipped with a Cryostream Coldhead 700 series. The temperature range was −150 °C to +70 °C with the heating speed 6 °C/min and step 5 or 10 °C. During measurement, the temperature was constant with ±1 °C fluctuations. The batch program for the temperature measurement was set in this way: (i) from 25 to 70 °C, step 5 °C and waiting 30 min at 70 °C, (ii) from 70 °C to −20 °C, step 5 °C, (iii) from −20 °C to −150 °C, step 10 °C, (iv) from −150 to 70 °C, step 10 °C, (v) cooling to 25 °C with cooling speed 6 °C/ min. 2.6. Differential Scanning Calorimetry. The measurement was done on a Setaram calorimeter, model DSC 131. The temperature range was 15−150 °C with the heating speed 10 °C/min.

3. RESULTS For trospium chloride, a crystal form was described by the Xray powder diffraction pattern soon after its discovery in 1967.34 Because of the difficulty of producing quality singlecrystals, TCl’s full crystal structure has never been solved. Initially, we encountered the same difficulty and so focused on the powder sample of trospium chloride by synchrotron diffraction. However, through intensive trial-and-error, we were ultimately able to prepare what we believe to be the first TCl crystals of sufficient quality for SXRD. 3.1. Synchrotron Based Powder Diffraction. The powder data of trospium chloride obtained from the synchrotron experiment were successfully indexed. The space group was determined as P21 with the following lattice parameters a = 9.1533(1) Å, b = 10.9261(2)Å, c = 11.0239(2) Å, β = 101.011(1)°. It was not possible to obtain a sensible model of TCl structure directly from the analysis of the powder data. The model of the trospium molecule was obtained by semiempirical quantum mechanical calculation (PM3, Hyperchem 8.035 software). Neither the positional nor the thermal parameters of any atoms were refined. The DASH36 structure refinements converged consistently to the same solution. Unfortunately, the R-factor was too high for the solution to be fully correct. It was clear that, in this case, synchrotron powder diffraction could not provide a good 5196



Article

The structures of TClP21 (Form I) and TCl P21/c (Form II) were very similar, having one TCl in the asymmetric unit. Both structures are almost exclusively held by hydrogen bonds (classical and nonclassical).The main feature in the crystal packing of both structures is the chlorine anion, which is the acceptor of the only classical hydrogen bond in the structure. The difference in the packing of the structures was seen in the weaker nonclassical hydrogen bonds. In the case of Form I, the chlorine anion was held by four nonclassical hydrogen bonds, with only one C−H−O bond saturating the free electron pair on the oxygen atom in the carbonyl group. In the case of Form II, the chlorine anion was held by five nonclassical hydrogen bonds, with two C−H−O bonds saturating the free electron pairs on the oxygen atoms in the hydroxyl and carbonyl groups. Both Forms I and II are substantially disordered. The asymmetrical part of the unit cell is composed of two partially occupied molecules. They are related by a noncrystallographic mirror plane perpendicular to the b axis, resulting in heavy overlaps (structures shown at the bottom of Scheme 1; Figure 1 in the Supporting Information). Altogether, three crystal forms of TCl were obtained from the ethanol mother liquor. Besides the above-mentioned polymorphs (Forms I and II), trospium chloride ethanol solvate (Form III) was described. The crystal structure of ethanol solvate has two molecules of trospium chloride and two molecules of ethanol in an asymmetric unit (Figure 5). The structure is held by a well-

explanation of the problem. After the single-crystal structure of Form I (Figure 2) was solved, the clear correspondence of Form I with the synchrotron data was confirmed. The original high R-factor obtained for a powder based only solution was a result of a disorder, which was impossible to detect by the powder data analysis. 3.2. Single-Crystal X-ray Diffraction. Initial attempts to grow a single-crystal suitable for SXRD were carried out from ethanol mother liquor. The majority of the crystals appeared to suffer from decomposition, as the center parts were nebulous; however, the outer parts were clear and, when analyzed separately, Form I (space group P21) was solved (Table 1, Figure 3). The structure was in good accordance with the synchrotron powder data.

Figure 3. Independent molecule of trospium chloride in Form I: Hbond shown as dotted line; disorder omitted for clarity.

However, super cell diffractions (Figure 4) were sometimes observed for clear single-crystals. Their structure was solved in the space group P21/c (with corresponding systematic extinction) and assigned as Form II. No reproducible procedure for the preparation of Form II was found. Screening attempts resulted in a series of samples with subtle differences in DSC, FTIR, and ssNMR.19

Figure 5. Asymmetric unit of TCl ethanol solvate (Form III): Hbonds shown as dotted lines; solvent molecules displayed in a ball-andstick mode.

developed system of classical and nonclassical H-bonds, with the contribution of edge-to-face π-interactions. The main motif of the structure is formed by a cation of trospium, a chloride anion and molecule of ethanol, which are held by strong O− H−Cl H-bonds. These motifs are bound together by nonclassical H-bonds, mostly C−H−O and C−H−Cl (Table 1 in the Supporting Information). 3.3. X-ray Powder Diffraction. Although some differences were found between the calculated powder patterns for Form I and Form II (arrows in Figure 6 indicate the peaks related to the doubling of one unit cell parameter), the X-ray powder patterns of Form I and Form II were very similar. It is noteworthy that, in the many cooling and evaporation crystallization samples from ethanol measured by XRPD, only Form I was detected (never Form II or III). If there was any

Figure 4. Observed diffractions (SXRD) in Ewald construction for Form I (left, lattice P21) and Form II (right, lattice P21/c). 5197



Article

Figure 6. Difference between calculated X-ray powder patterns of Forms I and II and of the experimentally aquired (laboratory diffractometer) sample of Form I.

admixture of forms other than Form I, it was below the detection limit of synchrotron powder diffraction. The only “bulk” sample we know to contain a certain amount of Form II in powder is the sample from slurry experiments to determine the thermodynamically stable polymorph done in nbutanol (Figure 7). However, only partial conversion from Form I to Form II was observed. Quantitative analysis of the mixture by the Rietveld refinement was not possible due to the high similarity between the crystal structures of Forms I and II. For further details see the Supporting Information. 3.4. Thermodynamical Stability. The thermodynamical stability and possible phase transformations of trospium chloride were investigated by DSC, temperature resolved XRPD, slurry determination of the stable polymorph, and, finally, temperature resolved SXRD. Two polymorphs of TCl were described, but we had difficulty preparing sufficient amounts of Form II, so in the first three methods, Form I was used. By DSC, no change was observed in the range 15−150 °C (see Supporting Information). The same result was obtained by temperature resolved XRPD in the range −150 to 70 °C (see Supporting Information). The only changes visible can be attributed to the expansion of the unit cell with increasing temperature. On the other hand, in slurry experiments at room temperature, a polymorphic transition from Form I to Form II was observed. Therefore, in this system, Form II is the thermodynamically stable polymorph. However, during the 6 weeks the slurry experiments took place, we detected only partial transformation in n-butanol and no transformation in methyl ethyl ketone. The transformation from Form I to Form II has quite low kinetics.

We also investigated the possibility of sample-handlingdependent transformations during single-crystal X-ray diffraction. Normally, we put the crystal in the stream of cold nitrogen (−133 °C), and it is cooled very rapidly. This usually lowers the chance at decomposition. We wanted to know if the cooling rate and also measurement temperature can cause phase transitions of TCl. Luckily, we were able to select diffraction quality single-crystals of both Forms I and II (from the same crystallization sample). Crystals were introduced to the difractometer at room temperature and were first cooled to −153 °C and then heated to 57 °C. Measurements of unit cell parameters were done every 20 °C. Changes in double cell peaks (“double” related to Form I only) were monitored for both Forms I and II. In crystals of both Forms I and II, no phase transition was observed during the whole treatment (see Supporting Information). Mainly based on slurry experiments, we can deduce that, out of Forms I and II, Form II is the more stable. However, we did not see a transition in any of the temperature based experiments mentioned in this paragraph. It is possible that this was caused by the slow kinetics of the transformation.

4. DISCUSSION In this section, we unravel the knotty problem of the polymorphism of trospium chloride: whether what we observed is actually polymorphism or just variations of the same structure caused by the crystallization conditions. First, we compare the structures of both polymorphs and highlight their similarities and differences. Next, we explain molecular disorder in the polymorph(s) of trospium chloride. Then, we show how that very molecular disorder is actually responsible for the polymorphs being strikingly similar, even if, 5198



Article

Figure 7. XRPD of the slurry sample from n-BuOH showing that the sample contained a certain amount of Form II. Final Le Bail fit showing the measured data (black thin-cross), calculated data (red line), and difference curve (blue line). Calculated Bragg positions are shown by vertical bars. Top: The unit cell of the P21 phase (Form I) was used to fit measured data. Green ellipses show unexplained reflections at 11.89° and 22.01° 2Θ positions. Bottom: The unit cell of the P21/c phase (Form II) was used to fit measured data. Reflections at 11.89° and 22.01° 2Θ positions are explained by (−1 1 1) and (0 2 3) reflections.

at first glance, they seem different (symmetry group, unit cell size). Next, we propose that the different spatial distributions of disordered molecules are the very thing that enables the existence of the two different polymorphs. We conclude our discussion by pondering whether, in this case, molecular disorder weakens or rather stabilizes the crystal structure.

4.1. Comparison of Structures. In the solved structures, we compared the conformations of the trospium molecule, as well as the crystal packing, in Forms I and II. The conformations of the trospium molecules were remarkably similar in both forms. (In both Forms I and II, the disordered molecules are identical, because they are 5199



Article

noncrystallographic symmetry related (x, −y, z + 1/2). So it is sensible to compare only the representative conformations of Forms I and II.) However, there was a very subtle difference between the conformations of the five-membered rings. We believe that this discrepancy was not actual but simply caused by problems with the structural refinement. In this part of the molecule, there was a strong overlap of the disorder components, making the exact determination of the atom positions impossible. Significant differences were evident when we compared the conformations of trospium in the polymorphs (Forms I and II) with those observed in the trospium chloride EtOH solvate (Form III). The most significant variation among the TCl molecules was in the rotation of the phenyl rings. The difference between the conformations of the Form III trospium molecules (pink and violet in Figure 8) also confirmed that the Figure 9. Fit of unit cells of Form I (orange) and Form II (lime); disordered molecules with lower occupancy omitted for clarity.

4.2. Molecular Disorder. The structures of both polymorphs (Forms I and II) were disordered. In Form I, the ratio of the disordered molecules was 57:43 [occupancy 0.5734(15)]; in Form II, it was 81:19 [occupancy 0.8090(8)]. In both forms, the observed disorder was generated by a pseudosymmetry mirror plane, x − y + 1/2z. Although a molecular disorder over a symmetry operation is not unheard of, it is still rare. Cases have been described of a disorder over a center of symmetry37 and over a 2-fold axis.38 However, to the best of our knowledge, no other researcher has reported molecular disorder over a mirror plane. The structures shown at the bottom of Scheme 1 show the two mirror images creating disorder. Trospium chloride has two chiral centers but is also symmetrical (Figure 1), making it achiral as a whole. If trospium chloride had true chirality, the red and blue molecules (structure shown at the bottom of Scheme 1; Figures 10−13) would correspond to two enantiomers in a racemic crystal lattice. It is important to understand that the structures shown at the bottom of Scheme 1 compare the conformations of the two mirror molecules of the trospium molecule without any artificial overlay; they show the exact disorder exhibited in Forms I and II. In Form I, the presence of the second mirror molecule was caused by a noncrystallographic symmetry operation. In Form II, both mirror molecules were generated by the c glide plane and, again, by the noncrystallographic symmetry. Consequently, even if the extent of the disorder in the polymorphs was different, the ratio between the two mirror molecules present in both polymorphs was 1:1(if we approximate the extent of the disorder in Form I to 50%, see Scheme 1).That is highly unusual because it means that the polymorphs contained the same molecules in the same conformations in the same ratio and even in the same packing. Thus, we had to ask ourselves this question: Do we really have two distinct polymorphs, or just variations of the same structure? 4.3. Distinct Polymorphs. According to Gavezzotti,39 when faced with a question like this, the deciding factor is the crystal symmetry. If the studied solid forms have different crystal symmetry, they are distinct polymorphs. The proof of the existence of two polymorphs of TCl with different crystal symmetry was acquired directly from the raw single-crystal diffraction data. By a simple transformation, we obtained the

Figure 8. Conformations of trospium in Form I (orange), Form II (lime), and Form III (pink and purple); for clarity, only representative molecules not related by any symmetry operation are displayed.

space group of this phase had been correctly determined as P1̅ with two molecules of TCl in the asymmetric unit. Initially, we had been uncertain because two unit cell angles of this structure were close to 90° (Table 1). But, because the molecules had different conformations, they could not be symmetry dependent, and thus, both had to be included in the independent part of the unit cell. The conformations of the trospium molecules were very similar in Forms I and II, but what about the packing? At first look, the dimensions of their unit cells seemed to differ (Table 1), but when overlaid over each other, they fit well (Figure 9) due to the symmetry relations between each other. Parameter a in Form I was almost exactly half of parameter c in Form II; c in Form I corresponded to a in Form II; b was the same for both. Angle β was very similar also, the difference being only 0.36°. Form II was packed slightly more densely than Form I; the difference between Form I and Form II in cell volume and calculated density (Table 1) was approximately 2%. 5200



Article

Scheme 1. Connection between Pseudosymmetrical Disorder Observed in Structures of Forms I and II and the Crystal Symmetry of Both Structures

Figure 11. Proposed spatial distribution of mirror molecules of trospium molecules enabling an overall crystal lattice with a group of symmetry P21: aproximately 50% disorder, domains.

Figure 10. Proposed spatial distribution of mirror molecules of trospium molecules enabling an overall crystal lattice with a group of symmetry P21: approximately 50% disorder; random distribution.

Figure 12. Proposed spatial distribution of trospium mirror molecules enabling an overall crystal lattice with a group of symmetry P21/c: approximately 20% disorder, random distribution.

observed diffractions in a square display called the Ewald construction (Figure 4). The diffractions were arranged in a grid equivalent to a crystal lattice in a reciprocal space. (Reciprocal space is a set of points that each correspond to a set of crystallographic planes, on which X-ray interferes.) It can be clearly seen that, because the polymorphs have different unit cell dimensions, they must also have a different number of molecules in each unit cell and a different group of symmetry. A

systematic extinction corresponding to different groups of symmetry was observed as well. Form I crystallized in a space group P21; Form II had the higher symmetry of P21/c. To recap, the polymorphs of TCl had different groups of symmetry, but they were very similar in packing and conformations, especially when the disorder is considered. Conversely, in most cases, polymorphs differ in packing and/or 5201



Article

the chloride anion and the edge-to-face interaction between the benzene rings of neighboring trospium molecules. The chloride anion was disordered in the same way as its corresponding trospium molecule (Figure 1 in the Supporting Information). Thus, this H-bond system was not disrupted. On the whole, in Form I, the disorder did not weaken the nonbonded interactions in any significant way. This was to be expected because if the disorder is omitted in Form II, its group symmetry results in the two mirror molecules alternating every other molecule. This alternating arrangement seems even more favorable energetically because Form II had a slightly firmer molecular packing, which resulted in a higher density. Also, Form II was determined to be more thermodynamically stable than Form I. On the other hand, even though Form II is the stable polymorph, Form I is by far the most common form of TCl when crystallized from ethanol. From this we can speculate that Form I is driven kinetically and Form II thermodynamically. Because Forms I and II differ basically only in the level of disorder, it is not surprising that the more organized Form II was found to be more stable.

Figure 13. Isosurfaces of trospium mirror molecules present in both polymorphs of TCl (Forms I and II).

conformation. How is it then possible that we observed two polymorphs and not just one? 4.4. Spatial Distributions. We propose two spatial distributions of trospium mirror molecules that resulted in the two observed polymorphs. We approximate that the ratio between the two mirror molecules of TCl was 1:1 in both polymorphs and that the unit cell of Form I was half the size of the Form II cell. From SXRD, we obtained the average unit cell based on the whole measured crystal. When disorder is present, there must be more than one type of unit cell in the crystal. To create models of the real structures of Forms I and II, each had to contain the two mirror molecules of TCl in a ratio of 1:1. However, they had to be arranged in different ways, so that the resulting overall lattice was different, having two molecules in a Form I unit cell and four molecules in a Form II unit cell (Figure 9). In Figures 10−12, the unit cell highlighted by the pink box is without any disorder. The “real” structure of Form I could be explained in two ways: random distribution (Figure 10) or domains (Figure 11). Surprisingly, when only cells containing the same mirror molecule were used to model Form I, the double-sized lattice of Form II could be observed in the prepared model. That is why we believe that the real structure of Form I also contains mixed cells (both mirror molecules in one cell), as is shown in Figure 10. Because Form II had a significantly lower disorder level (19%), the proposed “real” structure was made by simply exchanging a fifth of the pairs of the molecules of the same mirror molecule with the other. (By the term “pair” we mean pair of molecules that are related by a screw axis in Form II (mirror molecules of the same color in our figures.) In each case, the whole pair had to be exchanged so that the lattice with a bigger unit cell remained unchanged. 4.5. Effect of Disorder on Stability. As Habgood showed in the case of caffeine,7 molecules that are almost symmetrical are more prone to orientation disorder. In such cases, exchange of ordered to the disordered molecule does not cause any significant lattice disruption. This also applies to trospium chloride; the space occupied by the two mirror molecules of trospium chloride was virtually identical (Figure 13). In Forms I and II, the molecules were held together by a system of nonbonded interactions (Tables 1 and 2 in the Supporting Information), of which the most important were the hydrogen bond between the hydroxyl in the trospium and

5. CONCLUSION Three different crystal phases of trospium chloride were prepared and their structures solved: two polymorphs and one solvate. The crystal structures of both polymorphs were very similar and both exhibited pseudosymmetrical disorder, making them even more similar. The high degree of similarity between the crystal arrangements of both TCl polymorphs makes the study of this system more complex. In almost all cases, XRPD is able of easily identifying different polymorphs. However, the powder X-ray patterns of the polymorphs of trospium chloride were similar to the point that XRPD from a laboratory source was not able to distinguish between them. Were the polymorphs we discovered real or just the result of different growth conditions? The answer lay in the observed crystal symmetry. With the use of the single-crystal X-ray diffraction patterns, we showed that polymorphs of trospium chloride truly exhibit different overall symmetry. Form I crystallized in the space group P21 while Form II had the higher symmetry of P21/c. Therefore, the two described solid forms of trospium chloride can be designated as true polymorphs. And finally, even though Form I is by far the most common form of TCl when crystallized from ethanol, it was Form II that was discovered to be the stable polymorph.

■

ASSOCIATED CONTENT

S Supporting Information *

Crystallographic data files for Forms I, II, and III; data from temperature resolved XRPD of Form I; tables about hydrogen bonding; thermal ellipsoids and numbering of structures; a figure of the disordered TCl molecule in Form I; and the histograms from temperature resolved SXRD for Forms I and II. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Telephone: +420 220 444 200. E-mail: eliska.skorepova@ vscht.cz. Notes

The authors declare no competing financial interest. 5202



■

Article

(34) Bertholdt; Pfleger. Schulze Arzneim. Forsch. 1967, 17, 719−726. (35) Hyperchem 8.0; Hypercube, Inc.: 1115 NW 4th Street, Gainesville, Florida 32601, USA. (36) David, W. I. F.; Shankland, K.; van de Streek, J.; Pidcock, E.; Motherwell, W. D. S.; Cole, J. C. J. Appl. Crystallogr. 2006, 39, 910− 915. (37) Bar, I.; Bernstein, J. J. Phys. Chem. 1982, 86, 3223−3231. (38) Bar, I.; Bernstein, J. Acta Crystallograph., Sect. B 1981, 37, 569− 575. (39) Gavezzotti, A. J. Pharm. Sci. 2007, 96, 2232−2241.

ACKNOWLEDGMENTS The authors thank the Ministry of Education, Youth and Sports of the Czech Republic (research program MSM6046137302) and PraemiumAcademiae of the Academy of Sciences (ASCR) for its financial support. They also thank Michaela Chrastná for her initial work on the crystallization of trospium chloride.

■

REFERENCES

(1) Polymorphism in pharmaceutical solids (Drugs and the pharmaceutical sciences), 2nd ed.; Informa Healthcare: New York, 2009. (2) Storey, R. A.; Ymén, I. Solid State Characterization of Pharmaceuticals; John Wiley & Sons: 2011. (3) Hilfiker, R. Polymorphism; John Wiley & Sons: 2006. (4) Chopra, D.; Row, T. N. G. Cryst. Growth Des. 2008, 8, 848−853. (5) Byrn, S. R.; Pfeiffer, R. R.; Stephenson, G.; Grant, D. J. W.; Gleason, W. B. Chem. Mater. 1994, 6, 1148−1158. (6) Touzé, E.; Lorber, B.; Deniziak, M.; Becker, H. D.; Kern, D.; Giegé, R.; Sauter, C. Cryst. Growth Des. 2007, 7, 2195−2197. (7) Habgood, M. Cryst. Growth Des. 2011, 11, 3600−3608. (8) Chang, J.; Sandler, S. I. J. Chem. Phys. 2006, 125, 054705. (9) Vujovic, D.; Nassimbeni, L. R. Cryst. Growth Des. 2006, 6, 1595− 1597. (10) Clawson, J. S.; Kennedy-Gabb, S.; Lee, A. Y.; Copley, R. C. B. J. Pharm. Sci. 2011, 100, 4302−4316. (11) Copley, R. C. B.; Barnett, S. A.; Karamertzanis, P. G.; Harris, K. D. M.; Kariuki, B. M.; Xu, M.; Nickels, E. A.; Lancaster, R. W.; Price, S. L. Cryst. Growth Des. 2008, 8, 3474−3481. (12) Habgood, M.; Grau-Crespo, R.; Price, S. L. Phys. Chem. Chem. Phys. 2011, 13, 9590−9600. (13) Threlfall, T. L.; Gelbrich, T. Cryst. Growth Des. 2007, 7, 2297− 2297. (14) Murugan, N. A.; Sayeed, A. J. Chem. Phys. 2009, 130, 204514. (15) Dikundwar, A. G.; Pete, U. D.; Zade, C. M.; Bendre, R. S.; Guru Row, T. N. Cryst. Growth Des. 2012, 12, 4530−4534. (16) Delaney, S. P.; Pan, D.; Galella, M.; Yin, S. X.; Korter, T. M. Cryst. Growth Des. 2012, 12, 5017−5024. (17) Borodi, G.; Pop, M. M.; Onija, O.; Filip, X. Cryst. Growth Des. 2012, 12, 5846−5851. (18) Singh-Franco, D.; Machado, C.; Tuteja, S.; Zapantis, A. Clin. Ther. 2005, 27, 511−530. (19) Urbanova, M.; Sturcova, A.; Brus, J.; Benes, H.; Skorepova, E.; Kratochvil, B.; Cejka, J.; Sedenkova, I.; Kobera, L.; Policianova, O.; Sturc, A. J. Pharm. Sci. 2013, 102, 1235−1248. (20) Aberg, A. K. G. Methods for relaxation of smooth muscle contractions using Trospium. (21) Crystalline forms of trospium chloride. Disclosure Number IPCOM000205316D, http://ip.com/IPCOM/000205316, 2011. (22) Crystalline solvate form of trospium chloride. Disclosure Number IPCOM000197128, http://ip.com/IPCOM/000197128, 2010. (23) Scher, D.; Ryznal, R.; Blizzard, C. Complex of Trospium and Pharmaceutical Compositions Thereof. WO/2010/011813, January 29, 2010. (24) CrysAlisPro; Oxford Diffraction: 2002. (25) SIR92; Altomare, A.; Cascarano, G.; Giacovazzo, C.; Guagliardi, A.; Burla, M. C.; Polidori, G.; Camalli, M. J. Appl. Crystallogr. 1994, 27, 435. (26) Palatinus, L.; Chapuis, G. J. Appl. Crystallogr. 2007, 40, 786− 790. (27) CRYSTALS; Betteridge, P. W.; Carruthers, J. R.; Cooper, R. I.; Prout, K.; Watkin, D. J. J. Appl. Cryst. 2003, 36, 1487. (28) Petricek; Palatinus. Jana; Institute of Physics: Prague, The Czech Republic, 2006. (29) Spek, A. L. Acta Crystallogr., Sect. D 2009, 65, 148−155. (30) Cooper, R. I.; Gould, R. O.; Parsons, S.; Watkin, D. J. J. Appl. Crystallogr. 2002, 35, 168−174. (31) Mercury; CCDC: 2007. (32) Discovery Studio; Accelrys, Inc., 2013. (33) X’PERT HighScore Plus; PANanalytical BV: 2009. 5203


Trospium Chloride: Unusual Example of Polymorphism Based on

Recommend Documents