- Email: [email protected]
Quantitative and Molecular Analysis of Buspirone Hydrochloride Polymorphs
Quantitative and Molecular Analysis of Buspirone Hydrochloride Polymorphs M. SHEIKHZADEH, S. ROHANI, A. JUTAN, T. MANIFAR Department of Chemical and Biochemical Engineering, The University of Western Ontario, London, Ont., Canada N6A 5B9
Received 17 March 2006; revised 11 May 2006; accepted 14 June 2006 Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20723
ABSTRACT: Quantitative and molecular analysis have been performed on two main polymorphs of buspirone hydrochloride (BUS-HCl). Quantitative analysis of solid-state composition of pharmaceutical powders is necessary to ensure safety and efficacy of drug substance and to validate the production processes. In this study, X-ray powder diffraction and differential scanning calorimetry have been used for the quantitative analysis of two polymorphic forms of BUS-HCl. In addition, single crystal X-ray study of Form 1 revealed its crystal structure, however, a similar study on Form 2 was not possible due to the difficulties encountered in producing it. Molecular analysis including partial charge analysis has been performed by using Gaussian simulation package to find the electro-negativity pattern in molecule. 1H and 13C solid-state NMR spectra of polymorphs were recorded and also regression equations and QM theory were developed to predict the 1H and 13C NMR spectra through the use of atomic environmental descriptors. The NMR differences between the two polymorphs were discussed by using prediction and experimental results and description of each NMR shift for carbon and hydrogen atoms. ß 2006 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 96:569–583, 2007
Keywords: chloride
quantitative analysis; NMR; single crystal diffraction; buspirone hydro-
INTRODUCTION Buspirone hydrochloride (BUS-HCl) is a white crystalline water-soluble anti-anxiety drug with a molecular mass of 422. Chemically, BUS-HCl is N-[4-[4-(2-pyrimidinyl)-1-piperazinyl]butyl]-1, 1-cyclopentanediacetamide monohydrochloride. The molecular formula C21H31N5O2 HCl is represented by Scheme 1. BUS-HCl has several polymorphs including Form 1 with a melting point at 1888C and Form 2 with a melting point at 2038C. Close to 90% of BUSHCl produced in our experiments were either Forms 1 or 2. Based on thermal analysis and Correspondence to: Dr. S. Rohani (Telephone: 519-6614116; Fax: 519-661-3498; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 96, 569–583 (2007) ß 2006 Wiley-Liss, Inc. and the American Pharmacists Association
solubility measurements that have been performed by authors, these two crystal structure are enantiotrops and transformation temperature from Form 1 to Form 2 is 958C. In a recent study,1 qualitative characterization of Forms 1 and 2 polymorphs was reported using thermal and spectroscopy analyses as well as quantitative analysis using solid-state FTIR. In this study, various techniques such thermal analysis, X-ray powder diffraction (XRPD), and FTIR were used for quantitative analysis of the mixtures of Forms 1 and 2 of BUS-HCl. Different regression methods, chemometric techniques such as partial least square (PLS) and principal component regression (PCR) were used to develop calibration curves. In addition, solid-state 13C cross polarization magic angle spinning nuclear magnetic resonance (CPMAS-NMR) was used for investigating solid modifications of the drug substances. Because the
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
569
570
SHEIKHZADEH ET AL.
Scheme 1. Chemical structure of buspirone hydrochloride.
molecules in one polymorphic form of a drug are arrayed differently than those in another polymorphic form, differences in the solid-state spectra of the polymorphs are quite apparent and can be used for both qualitative and quantitative analysis of a polymorphic mixture. Chilmonczyk et al.2 has compared NMR and spectroscopic analysis of different buspirone analogous without considering the polymorphic form of this compound. Prediction of NMR is always used to find out the effect of different atomic groups on spectra.3 Also single crystal X-ray is one of the accurate methods to find the structure and differences between polymorphs.4 In this study, NMR estimation and single crystal studies were used to interpret differences of two polymorphic structures.
MATERIALS AND METHODS Buspirone freebase (BUS-base) was supplied by Apotex PharmaChem, Inc., (Brantford, ON). Other chemicals were purchased from Caledon (Georgetown, ON), and EMD (Gibbstown, NJ). Details on recrystallization of buspirone base and preparation of Forms 1 and 2 of BUS-HCl are reported by the same authors.1 Thermal Analysis Thermal analysis was done using a Mettler Toledo 822e DSC operating with version 6.1 Stare software. Samples of 4–15 mg were prepared in a covered aluminum crucible having pierced lids to allow escape of volatiles. The heating rate of 1, 5, 10, 20, 50, and 1008C/min were employed. The sensors and samples were under nitrogen purge JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
during the experiments and the flow rate of nitrogen was 30–50 mL/min. For quantitative analysis, uniform mixtures of two polymorphs were prepared. With 108C/min heating rate, thermograms of 7.33, 21.19, 49.79, 64.78, and 81.64% mixtures of Form 1 in Form 2 were obtained. The components were weighed to a total amount of 250 mg and mixed thoroughly and samples with a mass between 5 and 15 mg were prepared. Solid-state FTIR was used to make sure there was no transformation during the sample preparation. Data were processed by Stare version 6.1. Fourier Transform Infrared Spectroscopy (FTIR) The FTIR spectra were recorded on a solid-state Fourier transformation infrared spectrometer (Bruker Vector 22) equipped with OPUS v3.1. The samples were analyzed in transmission mode through a diamond window. The number of scans was 32 over the 450–4000/cm spectral region with a resolution of 2/cm. The background was collected in the same range for air. X-Ray Powder Diffraction (XRPD) The XRPD spectra were collected on a Rigaku– MiniFlex powder diffractometer, using CuKa (l ˚ ) radiation obtained at 30 kV for Ka ¼ 1.54059 A and 15 mA. The scans were run from 3.0 to 90.08 2y, increasing at a step size of 0.058 with a counting time of 1 s for each step. For quantitative analysis, both BUS-HCl forms were grinded to obtain uniform particles. Solid-state FTIR confirmed that there was no polymorphic transformation due to grinding. Powder diffractions of DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
7.33, 21.19, 49.79, 64.78, and 81.64% mixtures of Form 1 in Form 2 were recorded. The X-ray patterns for each sample were recorded over 3.0– 90.08 2y increasing at a step size of 0.058 with a counting time of 1 s for each step. The methods were validated for accuracy and precision. Instrument reproducibility was checked by recording the spectra of Form 1 three times without removing the sample from the XRPD machine. Repeatability was investigated at three powder compositions. Data were processed using the MDI-Jade version 7.5 software. Single Crystal Preparation For single crystal X-ray studies, attempts were made to produce single crystals of Form 1 and Form 2 using slow vaporization. In both cases the concentration of dissolved BUS-HCl in the different solvents was 0.5–2 wt% in 25 mL vials. Attempts to produce Form 2 single crystal failed. BUS-HCl Form 2 was dissolved in different solvents at different temperatures. Transformation from Form 2 to Form 1 occurs at around 958C. All attempts to produce Form 2 at temperatures higher than 958C in solvents such as xylene failed and only snow ball shape crystals with size of 10– 20 mm were produced. Form 1 single crystals were grown in methanol and acetonitrile. The vials were placed in an oven with slight vacuum for about 3–5 weeks. Form 1 samples were confirmed by solid-state FTIR before single crystal analysis. A crystal of BUS-HCl Form 1 was mounted on a glass fiber and data were collected at room temperature (208C) on a Nonius Kappa-CCD diffractometer with COLLECT (Nonius BV, 1998) and monochromatic Mo-Ka radiation ˚ ). The SHELXTL-NT V6.1 suite of (l ¼ 0.717 A programs was used to solve structure by direct methods. Computational Details Molecular energy, partial charge, dipole moment, and NMR prediction were determined using x86Linux-Gaussian 98 (RevA.9). Hartree–Fock theory using different basis sets including STO-G3, 3-21G, 6-31G(d), and 6-311þG(d,p) have been used. The multiplicity set to singlet and gasphase-optimized geometries were used in all cases. Calculation was conducted using computational chemistry laboratory of the Chemistry Department in the University of Western Ontario, London, ON. DOI 10.1002/jps
13
571
C and 1H NMR
All solid-state spectra were recorded on Bruker AV-300 spectrometer equipped with 4 mm broadband probe. The 13C-NMR spectra were recorded by CPMAS with spinning speed of 10 kHz. The 1 H-NMR solution spectrum in CDCl3 for checking the prediction results was performed on Bruker AV-600 spectrometer equipped with TB1 probe.
RESULTS AND DISCUSSION Quantitative Analysis In the case of sufficiently large differences between XRPD patterns and DSC thermographs of the two forms, quantification of the polymorphic mixtures was done using the following methods. X-Ray Powder Diffraction (XRPD) Quantitative analysis using XRPD data can be performed using different regression methods based on the analysis of a single peak, whole powder pattern, Chemometric, Smith, Toraya, or Rietveld methods. In this study, the intensity or area of a single peak was used for calibration and quantitative analysis.5 The characteristic peaks for Form 1 and Form 2 are at 8.5 and 16.48 2y, respectively (see Fig. 1). Using the intensity and area of these two peaks and linear regression, calibration curves of the mixtures of two BUS-HCl polymorphs were obtained. Figure 2a,b shows the linear regression for mixtures of two polymorphs using Form 1 and Form 2 characteristic peaks, respectively. It is noticed that the calibration curve using Form 2 characteristic peak gives better results. The area of each peak is also a useful tool for finding the calibration curve. Figure 3a,b shows the results for mixtures of two polymorphs using the area of the characteristic peaks. The calibration curve using area of peaks of From 1 gives better results rather than Form 2. Differential Scanning Calorimetry (DSC) Analysis The DSC analysis was used to confirm the polymorphic identity of the final product and any possible transformation during sample preparation. Sheikhzadeh et al.1 reported the DSC analysis for BUS-HCl polymorphs. With 18C/min heating rate, Form 1 exhibits one endotherm at JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
572
SHEIKHZADEH ET AL.
Figure 1. X-ray powder diffraction patterns of two polymorphs of BUS-HCl and their mixtures.
1898C, the melting point of Form 1, an exotherm at 1928C for crystallization to Form 2, and another endotherm at 2038C for melting of the recrystallized Form 2. Also Form 2 polymorph
Figure 2. XRPD calibration curves using intensity of characteristic peaks (a) Form 1 (2y ¼ 8.58), (b) Form 2 (2y ¼ 16.48). JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
Figure 3. XRPD calibration curves using areas of characteristic peaks (a) Form 1 (2y ¼ 8.58), (b) Form 2 (2y ¼ 16.48). DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
573
Figure 4. DSC results for quantitative analysis.
yielded a single melting point at 2038C and changing the heating rate did not make any effect on this thermogram. For the quantitative analysis, the heating rate used was 108C/min. Figure 4 shows the DSC results of known samples. The ratio of peak 1 to peak 2 heights and the area of peak 1 were used for calibration, respectively. Figure 5a,b presents the results of two calibration methods based on the area and ratio of peaks. The area of peak 2 gives better results. Table 1 lists and compares the quantitative methods used in this study with a recent study conducted by the same authors.1 Solid-state FTIR has been used to find calibration curve. Form 2 has special characteristic peak at 1156/cm and using the ratio between characteristic peak and one reference peak at 1193/cm, calibration curve was developed. Also, chemometric methods including PLS and PCR were implemented for different combinations of peaks. In Table 1, results for two calibration curves using four peak information and PLS and PCR methods have been compared with linear FTIR calibration curve and other DSC and XRPD results from this study.
mention anything about the polymorphic form of BUS-HCl and the results of this study proved that their crystal structure is very close to Form 1.
Molecular Analysis Single Crystal X-Ray Results Table 2 shows the results of the single crystal study of Form 1. The crystal data are fairly close to the work of Chilmonczyk et al.6 But they did not DOI 10.1002/jps
Figure 5. Calibration curves for mixture of two polymorphs based on DSC results. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
574
SHEIKHZADEH ET AL.
Table 1. Comparison of Different Quantitative Analysis (FTIR Result from Ref.1) Device
Equation (y is Form 1%)
‘‘x’’ Definition
R2
XRPD
y ¼ 0.2425x11.367 y ¼ 0.1555x þ 95.861 y ¼ 0.0208x37.395 y ¼ 0.0083x þ 98.127 y ¼ 118.16x þ 10.641 y ¼ 1.5884x þ 7.3161 y ¼ 237.09x þ 202.30 PCR PLS
Form 1 characteristic peak intensity Form 2 characteristic peak intensity Form 1 characteristic peak area Form 2 characteristic peak area Form 1/Form 2 (normalized peak area) Form 1 normalized peak area Form 2 characteristic peak/reference peak Method with 4 peaks (1118, 1134, 1153, and 1193/cm)
0.9097 0.9876 0.9437 0.9184 0.9046 0.9417 0.9866 0.9919 0.9823
DSC
FTIR
Symmetry transformation are (x,y,z),(2 1/2 x,1/ 2 þ y,1/2 z),(3 x,y, z) and (4 1/2 þ x, 1/ 2y, 1/2 þ z). Figure 6 shows one view of packing of Form 1 unit cell and hydrogen bonding associated with chloride ions. To find the position of H atom from hydrochloride acid and confirm the results of single crystal studies, partial charge, and dipole moment calculation were performed. Gaussian 98 software package was used for this calculation.7 Using different basis sets and the same HF theory, all the information for Form 1 has been calculated. Table 3 shows information from molecular modeling analysis. Different basis sets gave different result for the highest negative atom. STO-G3 and 6-21G are minimal and basic basis sets, respectively. But 6-31G(d) and 6-311 þ G(d,p) are routine and accurate basis sets for calculation
and their results are more reliable than STOG3 and 6-21G sets. Regarding to results from 6-31G(d) and 6-311þG(d,p), partial charge calculation revealed that among nitrogen and oxygen atoms as the main source to accept hydrogen atom (see Scheme 1), N[17] has the highest negative charge and hence the largest affinity to accept Hþ from HCl molecule. Single crystal structure studies demonstrated that the length of H. . .Cl and H. . .O bond are 4.89 ˚ , respectively. In addition the hydrogen and 2.20 A atom connected to N[17] can participate in forming intermolecular and intramolecular hydrogen bondings with the closest chloride ion and oxygen in the unit cell. Also to confirm the single crystal structure did not change or transform to another form during experiment, XRPD pattern prediction has been
Table 2. Crystal Structure Data of BUS-HCl from Chilmonczyk Work6 and This Study
Molecular formula Molecular weight Cell setting Space group Unit cell dimensions a b c a b g ˚ )3 Volume (A Z Density
This Study (BUS-HCl, Form 1)
Chilmonczyk et al., 19956
C21H32N5Oþ 2 , Cl 422 Monoclinic P 21/n
C21H32N5Oþ 2 , Cl 422 Monoclinic P 21/n
14.706 7.098 21.492 90 103.73 90 2179 4 1.27
14.747 7.101 21.536 90 103.79 90 2190.212 4 1.28
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
Figure 6. Unit cell structure for Form 1 from single crystal X-ray experiment.
575
These parameters are functions of other variables such as diffraction angle, X-ray beam properties, absorption coefficient. Using Jade-plus toolbox in Jade software (version 7.5), which is connected to XRPD machine that has been described in Section X-ray powder diffraction (XRPD), XRPD pattern simulation for single crystal structure data was evaluated. Calculated pattern was compared with BUSHCl Form 1 X-ray results. The sample for XRPD experiment was the same sample that was used for producing single crystal. Figure 7 shows the good agreement between XRPD result and simulated pattern from crystal structure data. This result proves that the single crystal structure belongs to Form 1 and no transformation or conversion occurred during the single crystal and X-ray experiment. NMR Analysis (Prediction of Experimental Results)
done. The XRPD pattern is presented by a list of possible reflections with their integrated intensities I (hkl) calculated using the equation: IðhklÞ ¼ KLp TAGmF 2
ð1Þ
where K is a scaling constant related to the intensity of X-ray beam. Since it is usually unknown, I (hkl) is normalized to the strongest reflection in the list and reported as relative I%. Lp is the Lorentz and polarization factor which is dependent on the diffraction geometry. T is the overall temperature factor of the structure. A is the absorption correction for flat specimen or zero background sample holder. G is the preferred orientation correction, m is the reflection multiplicity, and F is the structure factor derived from the packing of all atoms in the unit cell.
Solid-state NMR is a useful tool in investigating solid modifications of drug substances. In addition, because the molecules in one polymorphic form of a drug are arrayed differently compared to those in another polymorphic form, finding NMR shifts for all atoms in a molecule is quite informative. Prediction of chemical shifts in NMR is based on the effect of the constituent atoms in one molecule and usually the summation of these effects will give the NMR shift for that specific atom. With available rules in literature, it is possible to calculate the substructure effects. From experiment, differences between two polymorphs can be observed and predictive calculations give the interpretation for each atom and its effect on NMR shifts. Also quantum mechanics (QM) simulation of H and C NMR is another
Table 3. Partial Charge, Dipole Moment, and Energy Calculation with Using QM Analysis for Different Basis Sets Basis Set Atom O[10] O[12] N[7] N[17] N[20] N[24] N[28] Dipole moment (Debye) HF energy (Hartree)
DOI 10.1002/jps
HF/STO-3G
HF/3-21G
HF/6-31G(d)
HF/6-311 þ G(d,p)
0.398792 0.402365 0.457509 0.377724 0.374749 0.417489 0.417711 1.4916 1211.01
0.421442 0.424352 1.275193 0.901168 1.074946 0.670676 0.672702 1.7151 1220.88
0.354965 0.359978 0.030681 0.384828 0.185981 0.125286 0.124596 1.8909 1227.91
0.073216 0.060340 0.14407 0.26284 0.030125 0.100785 0.101245 2.0067 1228.37
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
576
SHEIKHZADEH ET AL.
Figure 7. Simulated XRPD pattern and Form1 XRPD pattern comparison.
accurate and reliable approach that was used for prediction. 13
C NMR
One of method for NMR prediction is based on linear combination of effects of other atomic groups. With information on the substructures and correction base, the following equation can be used to predict the NMR shift. XX i ¼ ai þ zi;j i ¼ 1; 2 . . . j ¼ ; ; ; ð2Þ i
j
where d is NMR shift for atom i, ai is correction base and z is the increment for a, b, d, g effects. In order to determine the increments for different kinds of atomic groups (substituents), empirical parameters and NMR prediction toolbox in ChemOffice 9 software package8 were used. Energy minimization with using MOPAC was applied on the molecule before prediction. Appendix A shows increments and correction base information for each carbon atom in BUS-HCl molecule. Linear combination of these increments gives the NMR shift prediction for carbon atoms. Shifts of the carbon atoms C[14], C[15], C[16], JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
C[18], and C[22] can be assigned from the effect of the protonated N[17] nitrogen atom. C[25], C[26], and C[27] are under the effect of pyrimidine ring and N[24] and N[28] nitrogen atoms. N[7], O[10], and O[11] have significant effects on C[9], C[11], C[13] and even C[6] and C[8] compared to other groups. Solid-state NMR experiments have been performed for two polymorphs with the same conditions. Figure 7 shows the results of the two polymorphic forms of BUS-HCl. QM analysis is another approach that can provide accurate source of information for molecular analysis. Energy calculation was performed for molecules using different basis sets from minimal to routine level (Fig. 8). Table 4 shows the differences between the observed and two predicted NMR results for the two polymorphs. Three high intensity peaks on spectra related to C – O, N2C ( – N), and CCN bonds are very close together and it seems that the effect of N or O on changing the carbon position on NMR spectra is not too strong. Carbons C[1], C[5], C[3], and C[16] are far from nitrogen or oxygen atoms. The differences between the two polymorphs are considerable, especially in the range between 20 and 30 ppm. DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
577
Figure 8. Solid-state 13C NMR spectra of two polymorphic forms of BUS-HCl.
Compared to four peaks for Form 1 in this range, Form 2 has just two peaks but the area of these peaks is larger than Form 1 peaks and it can present aliphatic group C[14] and C[15]. According to the predicted results, there should be one shift at 39.9 ppm, but for both polymorphs there is no peak at that point. The two high intensity peaks at 40.35 and 41.02 ppm for Form 1 and Form 2, respectively, can suggest that C[2], C[4], C[6], and C[8] have similar CCH2C structure in ring leading to the same shift NMR. QM prediction results using HF/6-31G theory give comparable NMR shifts to increment rule DOI 10.1002/jps
prediction. QM theory can estimate NMR shift for every atom but for the same group such as C[41] and C[42], shifts are close to each other and average of shifts was presented in Table 4. Figure 9 shows the comparison between the predicted (from increment rule) and actual values for both polymorphic forms. 1
H NMR Prediction of the Experimental Observations The same procedure as 13C NMR was applied for prediction of 1H NMR. Appendix B shows
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
578
SHEIKHZADEH ET AL.
Table 4. Comparison of Predicted and Observed Values of 13C NMR Shifts for Form 1 and Form 2 Predicted Shifts by QM Atom Number C[1], C[5] C[14] C[15] C[3] C[13] C[2], C[4] C[8], C[6] C[21], C[19] C[18], C[22] C[16] C[26] C[25], C[27] C[23] C[9], C[11]
Figure 9. Comparison of predicted values.
13
Predicted Using Increment Theory
HF 6-31G(d)
Form 1 Shifts
Form 2 Shifts
25 25.1 25.5 34.7 39.3 39.9 41.6 50 52.6 53.8 110.3 157.9 162.8 173.9
22.15 23.83 27.5 37.61 40.5 43.8 43.8 50.09 56.28 57.09 112.2 157.62 159.07 172.2
23.55 25.99 26.8 35.14 38.65 — 40.35 50.67 52.49 56.39 111.19 157.73 159.73 171.88
21.57 24.99 — 37.15 39.06 — 41.02 50.1 53.62 54.94 111.08 157.39 160.87 172.37
C NMR observed and
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
the increment and correction base information for all hydrogen atoms in the molecule. The solidstate 1H-NMR spectrum of BUS-HCl Form 1 showed no clear signals. But Form 2 produced much more intense signals. Figure 10 shows the result of 1H NMR for Form 2. The NMR spectra for Form 2 also did not give the desired expected results. Therefore, small quantities of Form 1 and Form 2 were dissolved in CDCl3 and the solution spectrum was recorded to confirm the prediction data. Polymorphic difference does not exist in the solution form but NMR analysis gives the results based on molecular structure. Figure 11 presents the results of NMR for BUSHCl in CDCl3. Solid-state NMR of Form 2 shows two shifts at 2.4 and 3.5 ppm and they are related to hydrogen atoms close to nitrogen group at pyrimidine ring and N(C – O)2 bonds. The second shift is the highest peak at CDCl3 NMR. According to Table 5, the prediction results using increment rules and QM method are acceptable in the case of peak positions and number of atoms (area under each curve). Figure 12 shows the comparison of the observed and predicted NMR shifts based on results from BUS-HCl in CDCl3. DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
DOI 10.1002/jps
Figure 10.
1
Figure 11.
1
579
H solid-state NMR spectra of BUS-HCl Form 2.
H NMR shifts of dissolved BUS-HCl on CDCl3. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
580
SHEIKHZADEH ET AL.
Table 5. Comparison of Predicted and Observed Values of 1H NMR Shifts for Form 1 and Form 2 Predicted Shifts by QM Atom Number H[31], H[32], H[37], H[38] H[45], H[46] H[30], H[35], H[29], H[36] H[43], H[44] H[33], H[34], H[39], H[40] H[47], H[48] H[49], H[59], H[52], H[53] H[41], H[42] H[50], H[51], H[54], H[55] — H[58] H[56], H[57]
Predicted Using Increment Theory
HF 6-31G(d)
Experiment Shifts
Integration
1.43 1.39 1.51 1.55 2.1 2.36 2.59 3.48 3.16 — 6.58 8.38
1.43 1.35 1.59 0.90 2.64 2.79 3.92 3.61 3.90 — 6.63 8.18
1.42 1.53 1.63 1.86 2.52 2.74 3 3.47 3.7 4.75 6.53 8.24
2.01 1.02 2.01 1.00 2.00 0.99 0.99 0.99 1.99 0.99 0.5 1.00
CONCLUSIONS There are numerous techniques and methods that allow quantification of solid-state forms of pharmaceuticals. It is difficult to predict which method is best suited for a particular type of a drug a priori. XRPD and DSC calibration curves for mixtures of two polymorphs of BUS-HCl were developed in this study based on their characteristic peak in XRPD and melting point peak in DSC. The results based on the intensity of the characteristic peak of Form 2 in XRPD and area of melting point peak of Form 1 in DSC had a smaller error than other methods. Single crystal X-ray revealed crystal structure of Form 1 and using molecular calculation based on HF theory and different basis sets from minimal to accurate level, molecular energy, partial charge, and dipole-moment were calculated. XRPD pattern
prediction was performed using From 1 structure and results confirmed the single crystal structure. Solid-state NMR analysis of hydrogen and carbon atoms was performed and remarkable differences between the two polymorphs were noted. Prediction of 1H and 13C NMR were conducted based on linear combination of increments and correction base information for each atom. Also using QM methods with different settings, NMR shifts for both polymorphs were evaluated. This prediction was useful to interpret NMR spectra to find specific atomic groups that have an effect on NMR spectra of each polymorph. Good agreement between 13C NMR and prediction results from increment rules and QM method suggested that the cyclopentane ring has a major effect on differences between two polymorphs of BUSHCl. Solid-state 1H NMR did not give good results because Form 1 did not show acceptable peaks and comparison between polymorphs was not feasible. 1H NMR prediction was conducted by two mentioned methods and results showed good agreement with experimental results from solidstate NMR for Form 2 and solution NMR for Form 1 and Form 2.
ACKNOWLEDGMENTS
Figure 12. Comparison of predicted values.
1
H NMR observed and
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
The authors thank ApotexPharmaChem Inc., Brantford, Ontario, Canada, for providing buspirone free base samples. Moreover, the financial support provided by NSERC CRD project and ApotexPharmaChem Inc. is greatly appreciated. DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
Appendix A.
13
581
C NMR Increments Information for BUS-HCl Molecule Substructure
Atom
Position
C[1], C[5]
Base a b g d d
C[2], C[4]
Base a b g g
C[3]
Base a b b b d
C[8], C[6]
Base a a b g g d
C[9], C[11]
C[13]
Base a a b b g d d
C[14]
Base a b b g g d
Cyclopentane C from aliphatic C from aliphatic C from aliphatic C(– O)N from aliphatic C– O from aliphatic General correction Cyclopentane C from aliphatic C from aliphatic C( –O)N from aliphatic C– O from aliphatic General correction Cyclopentane C from aliphatic C(– O)N from aliphatic C– O from aliphatic C from aliphatic C from aliphatic General correction Cyclohexane C( –O)N from aliphatic C from aliphatic C from aliphatic C– O from aliphatic C from aliphatic C from aliphatic General correction Amide CCC C– O from Namide C– C from Namide General correction Aliphatic C N C( –O)C C C C N General correction Aliphatic C C N C(– O)C N C
Number
Increment
Shift
1 2 2 2 1 1
11.4 18.2 18.8 5 0.4 0 4.8 11.4 18.2 37.6 3.2 2.7 1.4 11.4 36.4 2.6 0.6 18.8 0.3 11.4 2.3 22.5 9.1 28.2 2.7 7.5 0.3 6 165 11.5 2.0 1.8 1.2 2.3 9.1 28.3 1 9.4 2.5 0.3 0 4 2.3 18.2 9.4 11.3 5.4 5.1 0.6
25
1 2 4 1 1 1 4 1 1 2 1 1 1 1 3 1 3 1 1 1 1 1 1 1 1 2 1 1 1 1 1 2 1 1 2 1 2
39.9
34.7
41.6
173.9
39.3
25.1
(Continued)
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
582
SHEIKHZADEH ET AL.
Appendix A. (Continued) Substructure Atom
Position
Number
General correction C[15]
Base a b b g g d
Aliphatic C C N C N C
C[16]
Base a a b g
Aliphatic C N C C
C[18], C[22]
Base a a b b g g d
C[21], C[19]
Base a a b b b g d
C[23]
Base
C[25], C[27]
Base
C[26]
Base
1 2 1 1 2 1 2 General correction
General correction Cyclohexane C from aliphatic N from aliphatic C from aliphatic N from aliphatic Aliphatic chain C from aliphatic C from aliphatic General correction Cyclohexane C from aliphatic N from aliphatic Aliphatic chain C from aliphatic N from aliphatic C from aliphatic C from aliphatic General correction Pyrimidine N from 2—pyridine General correction Pyrimidine N from 2—pyridine General correction Pyrimidine N from 2—pyridine General correction
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
1 1 1 3 3 1 1 1 2 1 1 2 1 1 1 1 1 1 1 2 1 1 1 1 1 1 1
Increment 1.6 2.3 18.2 9.4 11.3 5 5.1 0.6 1.6 2.3 9.1 28.3 28.2 7.5 2 2.3 9.1 28.3 18.8 11.3 2.6 5 0.3 5.3 2.3 9.1 28.3 9.3 9.4 11.3 5 0.3 10.4 158 11.3 6.5 156.4 0.9 2.4 121.4 10.6 0.5
Shift 25.5
53.8
52.6
50
162.8
157.9
110.3
DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
Appendix B.
583
1
H NMR Increments Information for BUS-HCl Molecule Substructure
Atom H[45], H[46]
Increment
H[58]
Base b b Base b Base Base b b Base a b Base a b Base a b Base a b Base a b Base
H[56], H[57]
Base
H[31], H[32], H[37], H[38] H[30], H[35], H[29], H[36] H[43], H[44]
H[33], H[34], H[39], H[40]
H[47], H[48]
H[49], H[59], H[52], H[53]
H[50], H[51], H[54], H[55]
H[41], H[42]
Methylene C N(C)C Cyclopentane C from methylene Cyclopentane Methylene N(C– O)C– O C Methylene C( –O)N C Methylene N(C)C C Methylene N(C)C NC(R) Methylene NC(R) N(C)C Methylene N(C –O)C –O C Pyrimidine N from pyrimidine Pyrimidine N from pyrimidine General correction
REFERENCES 1. Sheikhzadeh M, Rohani S, Jutan A. Solid-state characterization of buspirone hydrochloride polymorphs. J Pharm Res 23:1043–1050. 2. Chilmonczyk Z, Cybulski J, Szelejeweska A, Les A. 1996. NMR studies of buspirone analogues. J Mol Struct 358:195–207. 3. Schaller RB, Munk ME, Pretsch E. 1996. Spectra estimation for computer-aided structure determination. J Chem Inf Compact Sci 36:239–243. 4. Lee DC, Webb M. 2003. Pharmaceutical Analysis. 9600 Garsington Road, Oxford, England OX4 2DQ: Blachwell Publication.
DOI 10.1002/jps
Number
Increment
Shift
1 1 1 1 2 1 1 1 1 1 1 3 1 1 1 1 1 1 1 1 1 1 1 1 2 1
1.37 0.04 0.06 1.43 0.08 1.51 1.37 0.22 0.04 1.37 0.85 0.12 1.37 1.03 0.04 1.37 1.03 0.19 1.37 1.73 0.06 1.37 2.15 0.04 7.36 0.78 8.78 0.31 0.09
1.39
1.43 1.51 1.55
2.1
2.36
2.59
3.16
3.48
6.58 8.38
5. Stephenson GA, Forbes RA, Reutzel-Edens SM. 2001. Characterization of the solid state: quantitative issue. Adv Drug Deliv Rev 48: 67–90. 6. Chilmonczyk Z, Les A, Wozniakowska A, Cybulski J, Kozio A, Gdaniec M. 1995. Buspirone analogues as ligands of the 5-HT1A receptor 1. The molecular structure of buspirone and two analogues. J Med Chem 38:1701–1710. 7. Gaussian 98. 2001. Software package. 8. MOPAC. 2005. Toolbox in ChemOffice Software package.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
Received 17 March 2006; revised 11 May 2006; accepted 14 June 2006 Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20723
ABSTRACT: Quantitative and molecular analysis have been performed on two main polymorphs of buspirone hydrochloride (BUS-HCl). Quantitative analysis of solid-state composition of pharmaceutical powders is necessary to ensure safety and efficacy of drug substance and to validate the production processes. In this study, X-ray powder diffraction and differential scanning calorimetry have been used for the quantitative analysis of two polymorphic forms of BUS-HCl. In addition, single crystal X-ray study of Form 1 revealed its crystal structure, however, a similar study on Form 2 was not possible due to the difficulties encountered in producing it. Molecular analysis including partial charge analysis has been performed by using Gaussian simulation package to find the electro-negativity pattern in molecule. 1H and 13C solid-state NMR spectra of polymorphs were recorded and also regression equations and QM theory were developed to predict the 1H and 13C NMR spectra through the use of atomic environmental descriptors. The NMR differences between the two polymorphs were discussed by using prediction and experimental results and description of each NMR shift for carbon and hydrogen atoms. ß 2006 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 96:569–583, 2007
Keywords: chloride
quantitative analysis; NMR; single crystal diffraction; buspirone hydro-
INTRODUCTION Buspirone hydrochloride (BUS-HCl) is a white crystalline water-soluble anti-anxiety drug with a molecular mass of 422. Chemically, BUS-HCl is N-[4-[4-(2-pyrimidinyl)-1-piperazinyl]butyl]-1, 1-cyclopentanediacetamide monohydrochloride. The molecular formula C21H31N5O2 HCl is represented by Scheme 1. BUS-HCl has several polymorphs including Form 1 with a melting point at 1888C and Form 2 with a melting point at 2038C. Close to 90% of BUSHCl produced in our experiments were either Forms 1 or 2. Based on thermal analysis and Correspondence to: Dr. S. Rohani (Telephone: 519-6614116; Fax: 519-661-3498; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 96, 569–583 (2007) ß 2006 Wiley-Liss, Inc. and the American Pharmacists Association
solubility measurements that have been performed by authors, these two crystal structure are enantiotrops and transformation temperature from Form 1 to Form 2 is 958C. In a recent study,1 qualitative characterization of Forms 1 and 2 polymorphs was reported using thermal and spectroscopy analyses as well as quantitative analysis using solid-state FTIR. In this study, various techniques such thermal analysis, X-ray powder diffraction (XRPD), and FTIR were used for quantitative analysis of the mixtures of Forms 1 and 2 of BUS-HCl. Different regression methods, chemometric techniques such as partial least square (PLS) and principal component regression (PCR) were used to develop calibration curves. In addition, solid-state 13C cross polarization magic angle spinning nuclear magnetic resonance (CPMAS-NMR) was used for investigating solid modifications of the drug substances. Because the
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
569
570
SHEIKHZADEH ET AL.
Scheme 1. Chemical structure of buspirone hydrochloride.
molecules in one polymorphic form of a drug are arrayed differently than those in another polymorphic form, differences in the solid-state spectra of the polymorphs are quite apparent and can be used for both qualitative and quantitative analysis of a polymorphic mixture. Chilmonczyk et al.2 has compared NMR and spectroscopic analysis of different buspirone analogous without considering the polymorphic form of this compound. Prediction of NMR is always used to find out the effect of different atomic groups on spectra.3 Also single crystal X-ray is one of the accurate methods to find the structure and differences between polymorphs.4 In this study, NMR estimation and single crystal studies were used to interpret differences of two polymorphic structures.
MATERIALS AND METHODS Buspirone freebase (BUS-base) was supplied by Apotex PharmaChem, Inc., (Brantford, ON). Other chemicals were purchased from Caledon (Georgetown, ON), and EMD (Gibbstown, NJ). Details on recrystallization of buspirone base and preparation of Forms 1 and 2 of BUS-HCl are reported by the same authors.1 Thermal Analysis Thermal analysis was done using a Mettler Toledo 822e DSC operating with version 6.1 Stare software. Samples of 4–15 mg were prepared in a covered aluminum crucible having pierced lids to allow escape of volatiles. The heating rate of 1, 5, 10, 20, 50, and 1008C/min were employed. The sensors and samples were under nitrogen purge JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
during the experiments and the flow rate of nitrogen was 30–50 mL/min. For quantitative analysis, uniform mixtures of two polymorphs were prepared. With 108C/min heating rate, thermograms of 7.33, 21.19, 49.79, 64.78, and 81.64% mixtures of Form 1 in Form 2 were obtained. The components were weighed to a total amount of 250 mg and mixed thoroughly and samples with a mass between 5 and 15 mg were prepared. Solid-state FTIR was used to make sure there was no transformation during the sample preparation. Data were processed by Stare version 6.1. Fourier Transform Infrared Spectroscopy (FTIR) The FTIR spectra were recorded on a solid-state Fourier transformation infrared spectrometer (Bruker Vector 22) equipped with OPUS v3.1. The samples were analyzed in transmission mode through a diamond window. The number of scans was 32 over the 450–4000/cm spectral region with a resolution of 2/cm. The background was collected in the same range for air. X-Ray Powder Diffraction (XRPD) The XRPD spectra were collected on a Rigaku– MiniFlex powder diffractometer, using CuKa (l ˚ ) radiation obtained at 30 kV for Ka ¼ 1.54059 A and 15 mA. The scans were run from 3.0 to 90.08 2y, increasing at a step size of 0.058 with a counting time of 1 s for each step. For quantitative analysis, both BUS-HCl forms were grinded to obtain uniform particles. Solid-state FTIR confirmed that there was no polymorphic transformation due to grinding. Powder diffractions of DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
7.33, 21.19, 49.79, 64.78, and 81.64% mixtures of Form 1 in Form 2 were recorded. The X-ray patterns for each sample were recorded over 3.0– 90.08 2y increasing at a step size of 0.058 with a counting time of 1 s for each step. The methods were validated for accuracy and precision. Instrument reproducibility was checked by recording the spectra of Form 1 three times without removing the sample from the XRPD machine. Repeatability was investigated at three powder compositions. Data were processed using the MDI-Jade version 7.5 software. Single Crystal Preparation For single crystal X-ray studies, attempts were made to produce single crystals of Form 1 and Form 2 using slow vaporization. In both cases the concentration of dissolved BUS-HCl in the different solvents was 0.5–2 wt% in 25 mL vials. Attempts to produce Form 2 single crystal failed. BUS-HCl Form 2 was dissolved in different solvents at different temperatures. Transformation from Form 2 to Form 1 occurs at around 958C. All attempts to produce Form 2 at temperatures higher than 958C in solvents such as xylene failed and only snow ball shape crystals with size of 10– 20 mm were produced. Form 1 single crystals were grown in methanol and acetonitrile. The vials were placed in an oven with slight vacuum for about 3–5 weeks. Form 1 samples were confirmed by solid-state FTIR before single crystal analysis. A crystal of BUS-HCl Form 1 was mounted on a glass fiber and data were collected at room temperature (208C) on a Nonius Kappa-CCD diffractometer with COLLECT (Nonius BV, 1998) and monochromatic Mo-Ka radiation ˚ ). The SHELXTL-NT V6.1 suite of (l ¼ 0.717 A programs was used to solve structure by direct methods. Computational Details Molecular energy, partial charge, dipole moment, and NMR prediction were determined using x86Linux-Gaussian 98 (RevA.9). Hartree–Fock theory using different basis sets including STO-G3, 3-21G, 6-31G(d), and 6-311þG(d,p) have been used. The multiplicity set to singlet and gasphase-optimized geometries were used in all cases. Calculation was conducted using computational chemistry laboratory of the Chemistry Department in the University of Western Ontario, London, ON. DOI 10.1002/jps
13
571
C and 1H NMR
All solid-state spectra were recorded on Bruker AV-300 spectrometer equipped with 4 mm broadband probe. The 13C-NMR spectra were recorded by CPMAS with spinning speed of 10 kHz. The 1 H-NMR solution spectrum in CDCl3 for checking the prediction results was performed on Bruker AV-600 spectrometer equipped with TB1 probe.
RESULTS AND DISCUSSION Quantitative Analysis In the case of sufficiently large differences between XRPD patterns and DSC thermographs of the two forms, quantification of the polymorphic mixtures was done using the following methods. X-Ray Powder Diffraction (XRPD) Quantitative analysis using XRPD data can be performed using different regression methods based on the analysis of a single peak, whole powder pattern, Chemometric, Smith, Toraya, or Rietveld methods. In this study, the intensity or area of a single peak was used for calibration and quantitative analysis.5 The characteristic peaks for Form 1 and Form 2 are at 8.5 and 16.48 2y, respectively (see Fig. 1). Using the intensity and area of these two peaks and linear regression, calibration curves of the mixtures of two BUS-HCl polymorphs were obtained. Figure 2a,b shows the linear regression for mixtures of two polymorphs using Form 1 and Form 2 characteristic peaks, respectively. It is noticed that the calibration curve using Form 2 characteristic peak gives better results. The area of each peak is also a useful tool for finding the calibration curve. Figure 3a,b shows the results for mixtures of two polymorphs using the area of the characteristic peaks. The calibration curve using area of peaks of From 1 gives better results rather than Form 2. Differential Scanning Calorimetry (DSC) Analysis The DSC analysis was used to confirm the polymorphic identity of the final product and any possible transformation during sample preparation. Sheikhzadeh et al.1 reported the DSC analysis for BUS-HCl polymorphs. With 18C/min heating rate, Form 1 exhibits one endotherm at JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
572
SHEIKHZADEH ET AL.
Figure 1. X-ray powder diffraction patterns of two polymorphs of BUS-HCl and their mixtures.
1898C, the melting point of Form 1, an exotherm at 1928C for crystallization to Form 2, and another endotherm at 2038C for melting of the recrystallized Form 2. Also Form 2 polymorph
Figure 2. XRPD calibration curves using intensity of characteristic peaks (a) Form 1 (2y ¼ 8.58), (b) Form 2 (2y ¼ 16.48). JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
Figure 3. XRPD calibration curves using areas of characteristic peaks (a) Form 1 (2y ¼ 8.58), (b) Form 2 (2y ¼ 16.48). DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
573
Figure 4. DSC results for quantitative analysis.
yielded a single melting point at 2038C and changing the heating rate did not make any effect on this thermogram. For the quantitative analysis, the heating rate used was 108C/min. Figure 4 shows the DSC results of known samples. The ratio of peak 1 to peak 2 heights and the area of peak 1 were used for calibration, respectively. Figure 5a,b presents the results of two calibration methods based on the area and ratio of peaks. The area of peak 2 gives better results. Table 1 lists and compares the quantitative methods used in this study with a recent study conducted by the same authors.1 Solid-state FTIR has been used to find calibration curve. Form 2 has special characteristic peak at 1156/cm and using the ratio between characteristic peak and one reference peak at 1193/cm, calibration curve was developed. Also, chemometric methods including PLS and PCR were implemented for different combinations of peaks. In Table 1, results for two calibration curves using four peak information and PLS and PCR methods have been compared with linear FTIR calibration curve and other DSC and XRPD results from this study.
mention anything about the polymorphic form of BUS-HCl and the results of this study proved that their crystal structure is very close to Form 1.
Molecular Analysis Single Crystal X-Ray Results Table 2 shows the results of the single crystal study of Form 1. The crystal data are fairly close to the work of Chilmonczyk et al.6 But they did not DOI 10.1002/jps
Figure 5. Calibration curves for mixture of two polymorphs based on DSC results. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
574
SHEIKHZADEH ET AL.
Table 1. Comparison of Different Quantitative Analysis (FTIR Result from Ref.1) Device
Equation (y is Form 1%)
‘‘x’’ Definition
R2
XRPD
y ¼ 0.2425x11.367 y ¼ 0.1555x þ 95.861 y ¼ 0.0208x37.395 y ¼ 0.0083x þ 98.127 y ¼ 118.16x þ 10.641 y ¼ 1.5884x þ 7.3161 y ¼ 237.09x þ 202.30 PCR PLS
Form 1 characteristic peak intensity Form 2 characteristic peak intensity Form 1 characteristic peak area Form 2 characteristic peak area Form 1/Form 2 (normalized peak area) Form 1 normalized peak area Form 2 characteristic peak/reference peak Method with 4 peaks (1118, 1134, 1153, and 1193/cm)
0.9097 0.9876 0.9437 0.9184 0.9046 0.9417 0.9866 0.9919 0.9823
DSC
FTIR
Symmetry transformation are (x,y,z),(2 1/2 x,1/ 2 þ y,1/2 z),(3 x,y, z) and (4 1/2 þ x, 1/ 2y, 1/2 þ z). Figure 6 shows one view of packing of Form 1 unit cell and hydrogen bonding associated with chloride ions. To find the position of H atom from hydrochloride acid and confirm the results of single crystal studies, partial charge, and dipole moment calculation were performed. Gaussian 98 software package was used for this calculation.7 Using different basis sets and the same HF theory, all the information for Form 1 has been calculated. Table 3 shows information from molecular modeling analysis. Different basis sets gave different result for the highest negative atom. STO-G3 and 6-21G are minimal and basic basis sets, respectively. But 6-31G(d) and 6-311 þ G(d,p) are routine and accurate basis sets for calculation
and their results are more reliable than STOG3 and 6-21G sets. Regarding to results from 6-31G(d) and 6-311þG(d,p), partial charge calculation revealed that among nitrogen and oxygen atoms as the main source to accept hydrogen atom (see Scheme 1), N[17] has the highest negative charge and hence the largest affinity to accept Hþ from HCl molecule. Single crystal structure studies demonstrated that the length of H. . .Cl and H. . .O bond are 4.89 ˚ , respectively. In addition the hydrogen and 2.20 A atom connected to N[17] can participate in forming intermolecular and intramolecular hydrogen bondings with the closest chloride ion and oxygen in the unit cell. Also to confirm the single crystal structure did not change or transform to another form during experiment, XRPD pattern prediction has been
Table 2. Crystal Structure Data of BUS-HCl from Chilmonczyk Work6 and This Study
Molecular formula Molecular weight Cell setting Space group Unit cell dimensions a b c a b g ˚ )3 Volume (A Z Density
This Study (BUS-HCl, Form 1)
Chilmonczyk et al., 19956
C21H32N5Oþ 2 , Cl 422 Monoclinic P 21/n
C21H32N5Oþ 2 , Cl 422 Monoclinic P 21/n
14.706 7.098 21.492 90 103.73 90 2179 4 1.27
14.747 7.101 21.536 90 103.79 90 2190.212 4 1.28
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
Figure 6. Unit cell structure for Form 1 from single crystal X-ray experiment.
575
These parameters are functions of other variables such as diffraction angle, X-ray beam properties, absorption coefficient. Using Jade-plus toolbox in Jade software (version 7.5), which is connected to XRPD machine that has been described in Section X-ray powder diffraction (XRPD), XRPD pattern simulation for single crystal structure data was evaluated. Calculated pattern was compared with BUSHCl Form 1 X-ray results. The sample for XRPD experiment was the same sample that was used for producing single crystal. Figure 7 shows the good agreement between XRPD result and simulated pattern from crystal structure data. This result proves that the single crystal structure belongs to Form 1 and no transformation or conversion occurred during the single crystal and X-ray experiment. NMR Analysis (Prediction of Experimental Results)
done. The XRPD pattern is presented by a list of possible reflections with their integrated intensities I (hkl) calculated using the equation: IðhklÞ ¼ KLp TAGmF 2
ð1Þ
where K is a scaling constant related to the intensity of X-ray beam. Since it is usually unknown, I (hkl) is normalized to the strongest reflection in the list and reported as relative I%. Lp is the Lorentz and polarization factor which is dependent on the diffraction geometry. T is the overall temperature factor of the structure. A is the absorption correction for flat specimen or zero background sample holder. G is the preferred orientation correction, m is the reflection multiplicity, and F is the structure factor derived from the packing of all atoms in the unit cell.
Solid-state NMR is a useful tool in investigating solid modifications of drug substances. In addition, because the molecules in one polymorphic form of a drug are arrayed differently compared to those in another polymorphic form, finding NMR shifts for all atoms in a molecule is quite informative. Prediction of chemical shifts in NMR is based on the effect of the constituent atoms in one molecule and usually the summation of these effects will give the NMR shift for that specific atom. With available rules in literature, it is possible to calculate the substructure effects. From experiment, differences between two polymorphs can be observed and predictive calculations give the interpretation for each atom and its effect on NMR shifts. Also quantum mechanics (QM) simulation of H and C NMR is another
Table 3. Partial Charge, Dipole Moment, and Energy Calculation with Using QM Analysis for Different Basis Sets Basis Set Atom O[10] O[12] N[7] N[17] N[20] N[24] N[28] Dipole moment (Debye) HF energy (Hartree)
DOI 10.1002/jps
HF/STO-3G
HF/3-21G
HF/6-31G(d)
HF/6-311 þ G(d,p)
0.398792 0.402365 0.457509 0.377724 0.374749 0.417489 0.417711 1.4916 1211.01
0.421442 0.424352 1.275193 0.901168 1.074946 0.670676 0.672702 1.7151 1220.88
0.354965 0.359978 0.030681 0.384828 0.185981 0.125286 0.124596 1.8909 1227.91
0.073216 0.060340 0.14407 0.26284 0.030125 0.100785 0.101245 2.0067 1228.37
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
576
SHEIKHZADEH ET AL.
Figure 7. Simulated XRPD pattern and Form1 XRPD pattern comparison.
accurate and reliable approach that was used for prediction. 13
C NMR
One of method for NMR prediction is based on linear combination of effects of other atomic groups. With information on the substructures and correction base, the following equation can be used to predict the NMR shift. XX i ¼ ai þ zi;j i ¼ 1; 2 . . . j ¼ ; ; ; ð2Þ i
j
where d is NMR shift for atom i, ai is correction base and z is the increment for a, b, d, g effects. In order to determine the increments for different kinds of atomic groups (substituents), empirical parameters and NMR prediction toolbox in ChemOffice 9 software package8 were used. Energy minimization with using MOPAC was applied on the molecule before prediction. Appendix A shows increments and correction base information for each carbon atom in BUS-HCl molecule. Linear combination of these increments gives the NMR shift prediction for carbon atoms. Shifts of the carbon atoms C[14], C[15], C[16], JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
C[18], and C[22] can be assigned from the effect of the protonated N[17] nitrogen atom. C[25], C[26], and C[27] are under the effect of pyrimidine ring and N[24] and N[28] nitrogen atoms. N[7], O[10], and O[11] have significant effects on C[9], C[11], C[13] and even C[6] and C[8] compared to other groups. Solid-state NMR experiments have been performed for two polymorphs with the same conditions. Figure 7 shows the results of the two polymorphic forms of BUS-HCl. QM analysis is another approach that can provide accurate source of information for molecular analysis. Energy calculation was performed for molecules using different basis sets from minimal to routine level (Fig. 8). Table 4 shows the differences between the observed and two predicted NMR results for the two polymorphs. Three high intensity peaks on spectra related to C – O, N2C ( – N), and CCN bonds are very close together and it seems that the effect of N or O on changing the carbon position on NMR spectra is not too strong. Carbons C[1], C[5], C[3], and C[16] are far from nitrogen or oxygen atoms. The differences between the two polymorphs are considerable, especially in the range between 20 and 30 ppm. DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
577
Figure 8. Solid-state 13C NMR spectra of two polymorphic forms of BUS-HCl.
Compared to four peaks for Form 1 in this range, Form 2 has just two peaks but the area of these peaks is larger than Form 1 peaks and it can present aliphatic group C[14] and C[15]. According to the predicted results, there should be one shift at 39.9 ppm, but for both polymorphs there is no peak at that point. The two high intensity peaks at 40.35 and 41.02 ppm for Form 1 and Form 2, respectively, can suggest that C[2], C[4], C[6], and C[8] have similar CCH2C structure in ring leading to the same shift NMR. QM prediction results using HF/6-31G theory give comparable NMR shifts to increment rule DOI 10.1002/jps
prediction. QM theory can estimate NMR shift for every atom but for the same group such as C[41] and C[42], shifts are close to each other and average of shifts was presented in Table 4. Figure 9 shows the comparison between the predicted (from increment rule) and actual values for both polymorphic forms. 1
H NMR Prediction of the Experimental Observations The same procedure as 13C NMR was applied for prediction of 1H NMR. Appendix B shows
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
578
SHEIKHZADEH ET AL.
Table 4. Comparison of Predicted and Observed Values of 13C NMR Shifts for Form 1 and Form 2 Predicted Shifts by QM Atom Number C[1], C[5] C[14] C[15] C[3] C[13] C[2], C[4] C[8], C[6] C[21], C[19] C[18], C[22] C[16] C[26] C[25], C[27] C[23] C[9], C[11]
Figure 9. Comparison of predicted values.
13
Predicted Using Increment Theory
HF 6-31G(d)
Form 1 Shifts
Form 2 Shifts
25 25.1 25.5 34.7 39.3 39.9 41.6 50 52.6 53.8 110.3 157.9 162.8 173.9
22.15 23.83 27.5 37.61 40.5 43.8 43.8 50.09 56.28 57.09 112.2 157.62 159.07 172.2
23.55 25.99 26.8 35.14 38.65 — 40.35 50.67 52.49 56.39 111.19 157.73 159.73 171.88
21.57 24.99 — 37.15 39.06 — 41.02 50.1 53.62 54.94 111.08 157.39 160.87 172.37
C NMR observed and
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
the increment and correction base information for all hydrogen atoms in the molecule. The solidstate 1H-NMR spectrum of BUS-HCl Form 1 showed no clear signals. But Form 2 produced much more intense signals. Figure 10 shows the result of 1H NMR for Form 2. The NMR spectra for Form 2 also did not give the desired expected results. Therefore, small quantities of Form 1 and Form 2 were dissolved in CDCl3 and the solution spectrum was recorded to confirm the prediction data. Polymorphic difference does not exist in the solution form but NMR analysis gives the results based on molecular structure. Figure 11 presents the results of NMR for BUSHCl in CDCl3. Solid-state NMR of Form 2 shows two shifts at 2.4 and 3.5 ppm and they are related to hydrogen atoms close to nitrogen group at pyrimidine ring and N(C – O)2 bonds. The second shift is the highest peak at CDCl3 NMR. According to Table 5, the prediction results using increment rules and QM method are acceptable in the case of peak positions and number of atoms (area under each curve). Figure 12 shows the comparison of the observed and predicted NMR shifts based on results from BUS-HCl in CDCl3. DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
DOI 10.1002/jps
Figure 10.
1
Figure 11.
1
579
H solid-state NMR spectra of BUS-HCl Form 2.
H NMR shifts of dissolved BUS-HCl on CDCl3. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
580
SHEIKHZADEH ET AL.
Table 5. Comparison of Predicted and Observed Values of 1H NMR Shifts for Form 1 and Form 2 Predicted Shifts by QM Atom Number H[31], H[32], H[37], H[38] H[45], H[46] H[30], H[35], H[29], H[36] H[43], H[44] H[33], H[34], H[39], H[40] H[47], H[48] H[49], H[59], H[52], H[53] H[41], H[42] H[50], H[51], H[54], H[55] — H[58] H[56], H[57]
Predicted Using Increment Theory
HF 6-31G(d)
Experiment Shifts
Integration
1.43 1.39 1.51 1.55 2.1 2.36 2.59 3.48 3.16 — 6.58 8.38
1.43 1.35 1.59 0.90 2.64 2.79 3.92 3.61 3.90 — 6.63 8.18
1.42 1.53 1.63 1.86 2.52 2.74 3 3.47 3.7 4.75 6.53 8.24
2.01 1.02 2.01 1.00 2.00 0.99 0.99 0.99 1.99 0.99 0.5 1.00
CONCLUSIONS There are numerous techniques and methods that allow quantification of solid-state forms of pharmaceuticals. It is difficult to predict which method is best suited for a particular type of a drug a priori. XRPD and DSC calibration curves for mixtures of two polymorphs of BUS-HCl were developed in this study based on their characteristic peak in XRPD and melting point peak in DSC. The results based on the intensity of the characteristic peak of Form 2 in XRPD and area of melting point peak of Form 1 in DSC had a smaller error than other methods. Single crystal X-ray revealed crystal structure of Form 1 and using molecular calculation based on HF theory and different basis sets from minimal to accurate level, molecular energy, partial charge, and dipole-moment were calculated. XRPD pattern
prediction was performed using From 1 structure and results confirmed the single crystal structure. Solid-state NMR analysis of hydrogen and carbon atoms was performed and remarkable differences between the two polymorphs were noted. Prediction of 1H and 13C NMR were conducted based on linear combination of increments and correction base information for each atom. Also using QM methods with different settings, NMR shifts for both polymorphs were evaluated. This prediction was useful to interpret NMR spectra to find specific atomic groups that have an effect on NMR spectra of each polymorph. Good agreement between 13C NMR and prediction results from increment rules and QM method suggested that the cyclopentane ring has a major effect on differences between two polymorphs of BUSHCl. Solid-state 1H NMR did not give good results because Form 1 did not show acceptable peaks and comparison between polymorphs was not feasible. 1H NMR prediction was conducted by two mentioned methods and results showed good agreement with experimental results from solidstate NMR for Form 2 and solution NMR for Form 1 and Form 2.
ACKNOWLEDGMENTS
Figure 12. Comparison of predicted values.
1
H NMR observed and
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
The authors thank ApotexPharmaChem Inc., Brantford, Ontario, Canada, for providing buspirone free base samples. Moreover, the financial support provided by NSERC CRD project and ApotexPharmaChem Inc. is greatly appreciated. DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
Appendix A.
13
581
C NMR Increments Information for BUS-HCl Molecule Substructure
Atom
Position
C[1], C[5]
Base a b g d d
C[2], C[4]
Base a b g g
C[3]
Base a b b b d
C[8], C[6]
Base a a b g g d
C[9], C[11]
C[13]
Base a a b b g d d
C[14]
Base a b b g g d
Cyclopentane C from aliphatic C from aliphatic C from aliphatic C(– O)N from aliphatic C– O from aliphatic General correction Cyclopentane C from aliphatic C from aliphatic C( –O)N from aliphatic C– O from aliphatic General correction Cyclopentane C from aliphatic C(– O)N from aliphatic C– O from aliphatic C from aliphatic C from aliphatic General correction Cyclohexane C( –O)N from aliphatic C from aliphatic C from aliphatic C– O from aliphatic C from aliphatic C from aliphatic General correction Amide CCC C– O from Namide C– C from Namide General correction Aliphatic C N C( –O)C C C C N General correction Aliphatic C C N C(– O)C N C
Number
Increment
Shift
1 2 2 2 1 1
11.4 18.2 18.8 5 0.4 0 4.8 11.4 18.2 37.6 3.2 2.7 1.4 11.4 36.4 2.6 0.6 18.8 0.3 11.4 2.3 22.5 9.1 28.2 2.7 7.5 0.3 6 165 11.5 2.0 1.8 1.2 2.3 9.1 28.3 1 9.4 2.5 0.3 0 4 2.3 18.2 9.4 11.3 5.4 5.1 0.6
25
1 2 4 1 1 1 4 1 1 2 1 1 1 1 3 1 3 1 1 1 1 1 1 1 1 2 1 1 1 1 1 2 1 1 2 1 2
39.9
34.7
41.6
173.9
39.3
25.1
(Continued)
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
582
SHEIKHZADEH ET AL.
Appendix A. (Continued) Substructure Atom
Position
Number
General correction C[15]
Base a b b g g d
Aliphatic C C N C N C
C[16]
Base a a b g
Aliphatic C N C C
C[18], C[22]
Base a a b b g g d
C[21], C[19]
Base a a b b b g d
C[23]
Base
C[25], C[27]
Base
C[26]
Base
1 2 1 1 2 1 2 General correction
General correction Cyclohexane C from aliphatic N from aliphatic C from aliphatic N from aliphatic Aliphatic chain C from aliphatic C from aliphatic General correction Cyclohexane C from aliphatic N from aliphatic Aliphatic chain C from aliphatic N from aliphatic C from aliphatic C from aliphatic General correction Pyrimidine N from 2—pyridine General correction Pyrimidine N from 2—pyridine General correction Pyrimidine N from 2—pyridine General correction
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007
1 1 1 3 3 1 1 1 2 1 1 2 1 1 1 1 1 1 1 2 1 1 1 1 1 1 1
Increment 1.6 2.3 18.2 9.4 11.3 5 5.1 0.6 1.6 2.3 9.1 28.3 28.2 7.5 2 2.3 9.1 28.3 18.8 11.3 2.6 5 0.3 5.3 2.3 9.1 28.3 9.3 9.4 11.3 5 0.3 10.4 158 11.3 6.5 156.4 0.9 2.4 121.4 10.6 0.5
Shift 25.5
53.8
52.6
50
162.8
157.9
110.3
DOI 10.1002/jps
BUSPIRONE HYDROCHLORIDE POLYMORPHS
Appendix B.
583
1
H NMR Increments Information for BUS-HCl Molecule Substructure
Atom H[45], H[46]
Increment
H[58]
Base b b Base b Base Base b b Base a b Base a b Base a b Base a b Base a b Base
H[56], H[57]
Base
H[31], H[32], H[37], H[38] H[30], H[35], H[29], H[36] H[43], H[44]
H[33], H[34], H[39], H[40]
H[47], H[48]
H[49], H[59], H[52], H[53]
H[50], H[51], H[54], H[55]
H[41], H[42]
Methylene C N(C)C Cyclopentane C from methylene Cyclopentane Methylene N(C– O)C– O C Methylene C( –O)N C Methylene N(C)C C Methylene N(C)C NC(R) Methylene NC(R) N(C)C Methylene N(C –O)C –O C Pyrimidine N from pyrimidine Pyrimidine N from pyrimidine General correction
REFERENCES 1. Sheikhzadeh M, Rohani S, Jutan A. Solid-state characterization of buspirone hydrochloride polymorphs. J Pharm Res 23:1043–1050. 2. Chilmonczyk Z, Cybulski J, Szelejeweska A, Les A. 1996. NMR studies of buspirone analogues. J Mol Struct 358:195–207. 3. Schaller RB, Munk ME, Pretsch E. 1996. Spectra estimation for computer-aided structure determination. J Chem Inf Compact Sci 36:239–243. 4. Lee DC, Webb M. 2003. Pharmaceutical Analysis. 9600 Garsington Road, Oxford, England OX4 2DQ: Blachwell Publication.
DOI 10.1002/jps
Number
Increment
Shift
1 1 1 1 2 1 1 1 1 1 1 3 1 1 1 1 1 1 1 1 1 1 1 1 2 1
1.37 0.04 0.06 1.43 0.08 1.51 1.37 0.22 0.04 1.37 0.85 0.12 1.37 1.03 0.04 1.37 1.03 0.19 1.37 1.73 0.06 1.37 2.15 0.04 7.36 0.78 8.78 0.31 0.09
1.39
1.43 1.51 1.55
2.1
2.36
2.59
3.16
3.48
6.58 8.38
5. Stephenson GA, Forbes RA, Reutzel-Edens SM. 2001. Characterization of the solid state: quantitative issue. Adv Drug Deliv Rev 48: 67–90. 6. Chilmonczyk Z, Les A, Wozniakowska A, Cybulski J, Kozio A, Gdaniec M. 1995. Buspirone analogues as ligands of the 5-HT1A receptor 1. The molecular structure of buspirone and two analogues. J Med Chem 38:1701–1710. 7. Gaussian 98. 2001. Software package. 8. MOPAC. 2005. Toolbox in ChemOffice Software package.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 3, MARCH 2007