- Email: [email protected]
Application of NIR Spectroscopy in Polymorphic Analysis: Study of Pseudo-Polymorphs Stability
Application of NIR Spectroscopy in Polymorphic Analysis: Study of Pseudo-Polymorphs Stability MARCELO BLANCO,1 DA´MARIH VALDE´S,1 ISIDRO LLORENTE,2 MIGUEL BAYOD2 1
Unidad de Quı´mica Analı´tica, Departamento de Quı´mica, Facultad de Ciencias, Universidad Auto´noma de Barcelona, Edificio Cn, 08193 Bellaterra, Barcelona, Spain 2
Departamento de Investigacio´n y Desarrollo, Astur Pharma, c/ Pen˜a Brava, Parcela 22B y 23, Polı´gono Industrial Silvota, 33192 Silvota, Asturias, Spain
Received 17 November 2004; revised 26 January 2005; accepted 1 March 2005 Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20362
ABSTRACT: The accelerated transformation of three azithromycin pseudo-polymorphs (viz. the anhydrous, monohydrate (MH), and dihydrate (DH) forms) at a high temperature and moisture level was examined by near infrared spectroscopy (NIRS). The most marked spectral differences between the pseudo-polymorphs occurred in the 1800–2200 nm region, which corresponds to the first overtone for water. The qualitative analysis of the NIR spectra for the pseudo-polymorphs following storage in a stove at 608C at 100% relative humidity for 60 days suggests that the crystalline forms (viz. the MH and DH) are stable, whereas the amorphous (anhydrous) form evolves to the DH. This was confirmed by determining the amounts of water and DH present in anhydrous azithromycin and the MH by use of partial least-squares regression (PLSR). The method used to quantify the DH in MH samples was developed and validated in accordance with the standards of the International Conference of Harmonization (ICH) and the European Medicines Agency (EMEA) with a view to its subsequent application by the pharmaceutical industry. The limits of detection (LD) and quantitation (LQ) for the DH in MH provided by the NIRS method were consistent with those obtained by X-ray diffraction (XRD) methodology. This testifies to the accuracy of the proposed method. ß 2005 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 94:1336–1342, 2005
Keywords: analysis; near-infrared spectroscopy; X-ray diffractometry; multivariate analysis; hydrates/solvates; Pseudo-polymorphism; azithromycin
INTRODUCTION Polymorphism is the ability of a solid substance to occur in two or more crystal habits with a different spatial arrangement in their molecules. A large number of solid pharmaceuticals are polymorphs. This has substantial effects on some properties of a high pharmaceutical interest including chemical stability, hygroscopicity, dissolution rate, vapor pressure, hardness, or surface tension, some of which facilitate handling Correspondence to: Marcelo Blanco (Telephone: (34) 93 5811367; Fax: (34) 93 5812379; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 94, 1336–1342 (2005) ß 2005 Wiley-Liss, Inc. and the American Pharmacists Association
1336
(e.g., tableting), and other important features such as bioavailability.1 Some polymorphic substances exhibit a disparate degree of solvation with the molecules of the solvent (water, alcohol) where they are crystallized; the forms in question, known as ‘‘solvates,’’ are strictly not polymorphs, but pseudo-polymorphs. Amorphous forms can be assumed to be pseudo-polymorphs. Polymorphs exhibit differences in thermodynamic stability: one form is usually more stable than the others, which will thus tend to evolve to the most stable— the process, however, is frequently very slow. Polymorphs in a pharmaceutical preparation can be partly or fully transformed into others during storage; this alters the bioavailability of the active principle concerned. The temperature and
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
POLYMORPHISM BY NIR
moisture levels in storage silos can favor the conversion of some polymorphic forms into others, which present different behavior (e.g., tableting, bioavailability, etc.). Likewise, wet grinding, press compaction, and the addition of excipients can facilitate polymorph interconversion during the manufacturing process. The technique of choice for the characterization of polymorphs, X-ray diffraction (XRD), is too slow for use in quality control analyses. The pharmaceutical industry therefore requires faster alternatives for this type of analysis. Near infrared spectroscopy (NIRS) meets many of the requirements for the ready, expeditious analysis of polymorphs as it allows one not only to characterize the different forms of an active principle, but also to determine the polymorphic purity of both the pure product and the final preparation. The NIRS technique has lately grown dramatically in use in the pharmaceutical field.2 In response to the interest aroused, the European Medicines Agency (EMEA)3 and the International Conference on Harmonization (ICH)4 have issued guidelines for the development and validation of NIRS analytical procedures for the pharmaceutical industry. As important as characterizing polymorphs or pseudo-polymorphs in pharmaceutical preparations is detecting potential transformations at the production, packaging, and storage stages. There has been some research in this direction including the study of polymorph transformations during the drying of glycine5 and theophylline,6,7 or the detection of the end-point of a polymorphic transformation.8,9 Azithromycin (C38H72N2O12) is a member of an antibiotic family related to erythromycin A. It occurs as three different pseudo-polymorphs, namely: an anhydrous form (A) and two crystalline forms (the monohydrate, MH and the dihydrate, DH). The DH is the most stable of the three. In previous work, a method for characterizing form A and quantifying DH in A was developed.10 The presence of these pseudo-polymorphs and of variable amounts of water of crystallization suggests a potential transformation between them under dry or high-moisture conditions. The aim of this work was to examine the stability of each azithromycin pseudo-polymorph under high temperature and moisture conditions by using the NIRS technique in order to study the possible transformation between the different forms that allow ensuring the presence of a single polymorph. The influence of the presence of DH in
1337
samples of A and MH on the stability of the polymorphs was also studied. Calibration models constructed by partial leastsquares regression (PLSR) were used to determine the DH and water contents in samples of A and MH. The proposed method for quantifying DH in MH samples was validated in accordance with the ICH and EMEA guidelines. Finally, the limits of detection (LD) and quantitation (LQ) were determined by XRD and compared with those provided by the NIRS technique.
EXPERIMENTAL Samples Samples of the three pseudo-polymorphs (viz. amorphous azithromycin, A, the MH, and the DH) from different production batches were characterized by XRD (Fig. 1). Calibration models were constructed from laboratory-made A–DH and MH–DH mixtures containing 0%–10% DH by using an AB204 analytical balance from Mettler–Toledo International, Inc. (Columbus, OH). The samples were blended by hand and their homogeneity checked from their NIR spectra: blending was stopped when two consecutive spectra were identical. The weighing data were used as reference values. The samples were stored either under the laboratory ambient conditions (1), in a desiccator containing P2O5(s) (2) or in a stove at 608C and 100% relative humidity (3), in order to examine the influence of the DH content, temperature and moisture. Pseudo-polymorph stability was studied by preparing three samples each of A and MH
Figure 1. X-ray diffraction (XRD) patterns for the three-polymorphs of azithromycin. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
1338
BLANCO ET AL.
containing variable proportions of DH (0%, 2% and 7%, respectively). The six samples were stored in a stove at 608C in two moisture conditions: dry medium and saturation medium (100% RH) for 60 days. Apparatus and Software Near infrared spectra over the wavelength range 1100–2500 nm were recorded in the transflectance mode on a FOSS NIRSystems 6500 spectrophotometer equipped with a rapid content analyzer (RCA) module. The instrument was governed and data processed by using the software Vision v. 2.21 from FOSS NIRS ystems (Laurel, MD). Samples were placed in glass cuvettes and turned over with a spatula prior to recording triplicate spectra (averages of 32 scans) over the range 1100–2500 nm. The average of each triplicate spectrum was subjected to various treatments. The reference spectrum was obtained from an empty glass cuvette. Partial least-squares regression (PLS1) and principal component analysis (PCA) were performed by using the software Unscrambler v. 7.8, from Camo Process (Trondheim, Norway). Calibration models were constructed by cross-validation, using the leave-one-out method; the number of PLS components used was that minimizing the variation of the residual variance with the number of factors. The best model was taken to be that resulting in the lowest relative standard errors of calibration (RSEC) and prediction (RSEP).10 Moisture in the samples was determined on a 716 DMS Titrino Karl–Fisher titrator from Metrohm (Buckingham, UK). XRD patterns were obtained on an X’Pert Philips difractometer, using a Cu tube (l ¼ ˚ ) at 50 kV at 40 mA, a graphite secondary 1.5418 A monochromator, Bragg–Brentano geometry and a scintillation detector. XRD patterns were recorded over the 2y range 5–258, using a step width of 0.038 and an irradiation time of 3 s in each step.
Table 1. Figures of Merit of the Univariate Calibration by XRD (2y ¼ 13.0558) for Determination of Dihydrate (DH) in MH Samples (MH) of Azithromycin Concentration Range Calibration line Detection limit Quantification limit
0%–5% X ¼ 345.5 (13.8) þ 37.9 (4.9) C r ¼ 0.992 LD ¼ 3 sa/b ¼ 0.5% LQ ¼ 10 sa/b ¼ 1.5%
X, counts; C, concentration of DH (wt%); , confidence interval; sa, standard deviation of the intercept; b, slope.
were used to estimate the LD and LQ for these XRD methods. Those for DH in A were reported in a previous paper10 and those for DH in MH are given in Table 1.
Stability Study Figure 2 shows the NIR spectra for the three azithromycin pseudo-polymorphs. The greatest differences occur in the region from 1800 to 2200 nm, where the strongest band for water (viz. its combination band) appears. The anhydrous form exhibits a weak band corresponding to adsorbed water, whereas the crystalline forms (MH and DH) exhibit stronger water bands with one and two absorbance maxima, respectively, corresponding to crystallization and adsorbed water. The crystalline forms, MH and DH, possess a theoretical water content of 2.3% and 4.6% (wt %), respectively. The slight hygroscopicity of MH and A raises their water content to 1%–3% (wt %) without altering the pseudo-polymorph.
RESULTS AND DISCUSSION The three azithromycin pseudo-polymorphs were characterized from the XRD patterns of Figure 1. Samples of amorphous azithromycin (A) and the MH containing known amounts of the DH were used to construct a univariate calibration model for the DH content in the other two pseudopolymorphs. The regression lines thus obtained JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
Figure 2. SNV-corrected absorbance spectra for samples of the three pseudo-polymorphs of azithromycin.
POLYMORPHISM BY NIR
1339
Figure 3. (a) Evolution of NIR spectra of amorphous sample (0% DH) over time. (b) Evolution of NIR spectra of monohydrate (MH) sample (0% DH) over time.
In order to determine the stability of the pseudopolymorphs under different conditions, samples of the three azithromycin forms were stored in a stove at 608C prior to recording their NIR spectra, which revealed that adsorbed water was lost from A and MH, but not from DH. The combined effect of temperature (608C) and moisture (100% RH) resulted in no alteration of DH—which can thus be deemed the most stable form of the three. On the other hand, it caused A and MH to adsorb moisture. Figure 3 shows the NIR spectra in the region 1800–2200 nm for a sample of A and one of MH that were subjected to the above-described conditions for 60 days. As can be seen, both pseudopolymorphs present different behaviors. First the form A shows an increase of the water band, follow by an increase of the band ca. 1955 nm that corresponds to the DH form, therefore the form A structure absorbed water, that facilitate the conversion into DH. The MH only exhibited an increase in the first water band during the first 24 h, after which it leveled off; therefore, MH initially adsorbed an amount of moisture that remained constant thereafter, so it underwent no polymorphic transformation and was thus stable form under these conditions. The influence of the presence of DH (2% and 7%) in A and MH was studied in samples subjected to the same temperature and moisture conditions as the pure pseudo-polymorphs. The presence of DH in the MH had no effect on the stability of this pseudo-polymorph. On the other hand, the rate of the polymorphic transformation of A into DH increased with increasing DH content (A þ 7%
DH > A þ 2% DH > A), so the presence of DH in A accelerates its polymorphic transformation. Quantitative Analysis (PLS Models) The qualitative analysis exposed the behavior of A and MH under the conditions used in the stability study. In this study we apply the PLS1 calibration models to quantify the DH and water contents in samples of A and MH. However, in previous work, a method for determining DH and moisture contents of 0%–10% in samples of A was developed.10 The models examined in this work had to be adjusted to span the content range 0%–100%, however. The best results were obtained from second-derivative spectra in the region 1800– 2200 nm and three PLS factors. The DH and moisture contents in samples of MH were quantified by using PLS1 models involving different wavelength ranges (1100–2500 nm and 1800–2200 nm) spectral pretreatments (SNV, first and second derivative) and numbers of factors. The best results for both analytes (i.e., those exhibiting the lowest percent RSEC and RSEP values) are shown in Table 2. The previous PLS1 calibration models were used to determine DH and moisture in samples of A and MH. Figure 4 shows the predicted amounts of DH and water in samples of A containing 0%, 2%, or 7% DH. As can clearly bee seen, the contents in both moisture and DH increased with time, which confirms the occurrence of a polymorphic transformation. The DH contents of MH samples containing 0%, 2%, or 7% DH predicted by the proposed model JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
1340
BLANCO ET AL.
Table 2. Salient Features of the PLS Models Used in Determination of Dihydrate (DH) and H2O in Monohydrate Samples (MH) of Azithromycin Model Wavelength range (nm) Spectral pretreatment Concentration range PLS-factors Calibration
Prediction
DH
H2O
1100–2500 2nd derivative 0%–13.3% 4 n ¼ 39 CNIR ¼ 0.0691 þ 0.9863 CREF r ¼ 0.993 RSEC (%) ¼ 6.8 RMSEC (%) ¼ 0.4 n ¼ 44 CNIR ¼ 0.0584 þ 0.9885 CREF r ¼ 0.989 RSEP (%) ¼ 7.8 RMSEP (%) ¼ 0.4
1800–2200 2nd derivative 2.9%–5.4% 2 n ¼ 33 CNIR ¼ 0.0972 þ 0.9771 CREF r ¼ 0.988 RSEC (%) ¼ 2.5 RMSEC (%) ¼ 0.1 n ¼ 26 CNIR ¼ 0.1244 þ 1.0287 CREF r ¼ 0.977 RSEP (%) ¼ 2.7 RMSEP (%) ¼ 0.1
C, concentration (wt%).
Figure 4. (a) Prediction of DH contents in amporphous samples. (b) Prediction of moisture contents in amorphous samples.
remained constant with time, which confirms that this pseudo-polymorph is stable. As can be seen in Figure 5, the weight of the samples increased roughly by 1% (adsorbed water) within the first 24 h, after which it leveled off. Based on both the qualitative and quantitative results, MH is stable and A is transformed into DH under high temperature and moisture conditions.
Validation The proposed analytical method for determining the DH content in MH samples by NIRS was validated with a view to its use by the JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
Figure 5. Prediction of moisture in MH samples.
POLYMORPHISM BY NIR
1341
Table 3. Results of the Validation for the Proposed Method for the Quantitation of Dihydrate in Monohydrate Samples Parameter Linearity
Accuracy
Repeatability
Intermediate precision
Detection limit Quantification limit
Procedure
Results
CNIR ¼ a þ b CREF n¼8 Range ¼ 0%–10% Paired t-test of difference between NIR and reference value (n ¼ 8)
a ¼ 0.19 0.44 b ¼ 1.03 0.08 r ¼ 0.996 Residuals average ¼ 0.05 s ¼ 0.31 texp ¼ 0.459 ttab(a ¼ 0.05) ¼ 2.365 x1 ¼ 3.1 x2 ¼ 4.4 x3 ¼ 6.8
Three samples in three concentrations level analyzed by the same operator. Calculation of percent CV Sample analyzed 3 days by two different operators. ANOVA and calculation of percent CV LD ¼ 3.sa/b LQ ¼ 10.sa/b
x ¼ 4.4
CV1 ¼ 2.7 CV2 ¼ 4.4 CV3 ¼ 3.5 CVtotal ¼ 4.0
Not differences between days and operators
CV ¼ 4.8 LD ¼ 0.2% LQ ¼ 0.8%
C, concentration of DH (wt%); , confidence interval; sa, standard deviation of the intercept; b, slope.
pharmaceutical industry in accordance with the International Conference of Harmonization (ICH) and the EMEA guidelines. This purpose must be considered in determining the validation parameters required (viz. selectivity, linearity, range, accuracy, precision—as repeatability, and intermediate precision—, and LD and LQ).
the characteristics from the validation procedure for each one parameter and they results. The LD and LQ are equivalent as the obtained by XRD and testify to the suitability of the method for the intended purpose.
CONCLUSIONS Selectivity We addressed the identification of MH and DH azithromycin. We constructed a library from spectra for pure MH and DH azithromycin from different production batches (n ¼ 48). We used the 2nd derivative spectral mode, the wavelength range of 1100–2500 nm and the correlation coefficients in wavelength space as discriminating parameter, with a threshold of 0.90. We assessed the selectivity of the identification library by using it to identify external samples of MH and DH azithromycin. All samples were correct identified.
Quantitative Parameters The results for the validation of quantitative parameters are shown in Table 3, that indicate
The use of the NIRS technique to study the accelerated transformation of azithromycin pseudo-polymorphs under accelerated temperature and moisture conditions (at 608C and 100% RH) revealed that the two crystalline forms (viz. the MH and DH) are stable, whereas the amorphous form evolves to the DH, which is the most stable of the three pseudo-polymorphs. The proposed method for quantifying the DH in MH samples testifies to the suitability of NIRS methodology for characterizing and determining azithromycin pseudo-polymorphs.
ACKNOWLEDGMENTS The authors are grateful to Spain’s Ministry of Science and Technology (MCyT) for funding JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
1342
BLANCO ET AL.
this research within the framework or Project BQU2003-04247.
REFERENCES 1. Grant DJW. 1999. Theory and origin of polymorphism. In: Brittain HG, editor. Polymorphism in pharmaceutical solids, edn. New York: Marcel Dekker, Inc. pp 1–10. 2. Bakeev KA. 2003. Near-infrared spectroscopy as a process analytical tool. Pharm Technol Europe 15:27–28, 30–32. 3. Note for guidance on the use of near infrared spectroscopy by the pharmaceutical industry. 2003. http://www.emea.eu.int/pdfs/human/qwp/ 330901en.pdf 4. ICH quality guidelines. Q2A, text on validation of analytical procedures 1994; Q2B, validation of analytical procedures: methodology. 1996. http:// www.ich.org 5. Davis TD, Peck GE, Stowell JG, Morris KR, Byrn SR. 2004. Modeling and monitoring of polymorphic transformations during the drying phase of wet granulation. Pharm Res 21:860–866.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
6. Rasanen E, Rantanen J, Jorgensen A, Karjalainen M, Paakkari T, Yliruusi J. 2001. Novel identification of pseudopolymorphic changes of theophylline during wet granulation using near infrared spectroscopy. J Pharm Sci 90:389–396. 7. Airaksinen S, Karjalainen M, Raesaenen E, Rantanen J, Yliruusi J. 2004. Comparison of the effects of two drying methods on polymorphism of theophylline. Int J Pharm 276:129–141. 8. Norris T, Aldridge PK, Sekulic SS. 1997. Determination of end-points for polymorph conversions of crystalline organic compounds using online nearinfrared spectroscopy. Analyst (Cambridge, United Kingdom) 122:549–552. 9. De Braekeleer K, De Maesschalck R, Hailey PA, Sharp DCA, Massart DL. 1999. Online application of the orthogonal projection approach (OPA) and the soft independent modeling of class analogy approach (SIMCA) for the detection of the end point of a polymorph conversion reaction by near infrared spectroscopy (NIR). Chemometrics and Intelligent Laboratory Systems 46:103–116. 10. Blanco M, Valde´s D, Bayod MS, Ferna´ndez-Mari F, Llorente I. 2004. Characterization and analysis of polymorphs by near-infrared spectrometry. Analytica Chimica Acta 502:221–227.
Unidad de Quı´mica Analı´tica, Departamento de Quı´mica, Facultad de Ciencias, Universidad Auto´noma de Barcelona, Edificio Cn, 08193 Bellaterra, Barcelona, Spain 2
Departamento de Investigacio´n y Desarrollo, Astur Pharma, c/ Pen˜a Brava, Parcela 22B y 23, Polı´gono Industrial Silvota, 33192 Silvota, Asturias, Spain
Received 17 November 2004; revised 26 January 2005; accepted 1 March 2005 Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20362
ABSTRACT: The accelerated transformation of three azithromycin pseudo-polymorphs (viz. the anhydrous, monohydrate (MH), and dihydrate (DH) forms) at a high temperature and moisture level was examined by near infrared spectroscopy (NIRS). The most marked spectral differences between the pseudo-polymorphs occurred in the 1800–2200 nm region, which corresponds to the first overtone for water. The qualitative analysis of the NIR spectra for the pseudo-polymorphs following storage in a stove at 608C at 100% relative humidity for 60 days suggests that the crystalline forms (viz. the MH and DH) are stable, whereas the amorphous (anhydrous) form evolves to the DH. This was confirmed by determining the amounts of water and DH present in anhydrous azithromycin and the MH by use of partial least-squares regression (PLSR). The method used to quantify the DH in MH samples was developed and validated in accordance with the standards of the International Conference of Harmonization (ICH) and the European Medicines Agency (EMEA) with a view to its subsequent application by the pharmaceutical industry. The limits of detection (LD) and quantitation (LQ) for the DH in MH provided by the NIRS method were consistent with those obtained by X-ray diffraction (XRD) methodology. This testifies to the accuracy of the proposed method. ß 2005 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 94:1336–1342, 2005
Keywords: analysis; near-infrared spectroscopy; X-ray diffractometry; multivariate analysis; hydrates/solvates; Pseudo-polymorphism; azithromycin
INTRODUCTION Polymorphism is the ability of a solid substance to occur in two or more crystal habits with a different spatial arrangement in their molecules. A large number of solid pharmaceuticals are polymorphs. This has substantial effects on some properties of a high pharmaceutical interest including chemical stability, hygroscopicity, dissolution rate, vapor pressure, hardness, or surface tension, some of which facilitate handling Correspondence to: Marcelo Blanco (Telephone: (34) 93 5811367; Fax: (34) 93 5812379; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 94, 1336–1342 (2005) ß 2005 Wiley-Liss, Inc. and the American Pharmacists Association
1336
(e.g., tableting), and other important features such as bioavailability.1 Some polymorphic substances exhibit a disparate degree of solvation with the molecules of the solvent (water, alcohol) where they are crystallized; the forms in question, known as ‘‘solvates,’’ are strictly not polymorphs, but pseudo-polymorphs. Amorphous forms can be assumed to be pseudo-polymorphs. Polymorphs exhibit differences in thermodynamic stability: one form is usually more stable than the others, which will thus tend to evolve to the most stable— the process, however, is frequently very slow. Polymorphs in a pharmaceutical preparation can be partly or fully transformed into others during storage; this alters the bioavailability of the active principle concerned. The temperature and
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
POLYMORPHISM BY NIR
moisture levels in storage silos can favor the conversion of some polymorphic forms into others, which present different behavior (e.g., tableting, bioavailability, etc.). Likewise, wet grinding, press compaction, and the addition of excipients can facilitate polymorph interconversion during the manufacturing process. The technique of choice for the characterization of polymorphs, X-ray diffraction (XRD), is too slow for use in quality control analyses. The pharmaceutical industry therefore requires faster alternatives for this type of analysis. Near infrared spectroscopy (NIRS) meets many of the requirements for the ready, expeditious analysis of polymorphs as it allows one not only to characterize the different forms of an active principle, but also to determine the polymorphic purity of both the pure product and the final preparation. The NIRS technique has lately grown dramatically in use in the pharmaceutical field.2 In response to the interest aroused, the European Medicines Agency (EMEA)3 and the International Conference on Harmonization (ICH)4 have issued guidelines for the development and validation of NIRS analytical procedures for the pharmaceutical industry. As important as characterizing polymorphs or pseudo-polymorphs in pharmaceutical preparations is detecting potential transformations at the production, packaging, and storage stages. There has been some research in this direction including the study of polymorph transformations during the drying of glycine5 and theophylline,6,7 or the detection of the end-point of a polymorphic transformation.8,9 Azithromycin (C38H72N2O12) is a member of an antibiotic family related to erythromycin A. It occurs as three different pseudo-polymorphs, namely: an anhydrous form (A) and two crystalline forms (the monohydrate, MH and the dihydrate, DH). The DH is the most stable of the three. In previous work, a method for characterizing form A and quantifying DH in A was developed.10 The presence of these pseudo-polymorphs and of variable amounts of water of crystallization suggests a potential transformation between them under dry or high-moisture conditions. The aim of this work was to examine the stability of each azithromycin pseudo-polymorph under high temperature and moisture conditions by using the NIRS technique in order to study the possible transformation between the different forms that allow ensuring the presence of a single polymorph. The influence of the presence of DH in
1337
samples of A and MH on the stability of the polymorphs was also studied. Calibration models constructed by partial leastsquares regression (PLSR) were used to determine the DH and water contents in samples of A and MH. The proposed method for quantifying DH in MH samples was validated in accordance with the ICH and EMEA guidelines. Finally, the limits of detection (LD) and quantitation (LQ) were determined by XRD and compared with those provided by the NIRS technique.
EXPERIMENTAL Samples Samples of the three pseudo-polymorphs (viz. amorphous azithromycin, A, the MH, and the DH) from different production batches were characterized by XRD (Fig. 1). Calibration models were constructed from laboratory-made A–DH and MH–DH mixtures containing 0%–10% DH by using an AB204 analytical balance from Mettler–Toledo International, Inc. (Columbus, OH). The samples were blended by hand and their homogeneity checked from their NIR spectra: blending was stopped when two consecutive spectra were identical. The weighing data were used as reference values. The samples were stored either under the laboratory ambient conditions (1), in a desiccator containing P2O5(s) (2) or in a stove at 608C and 100% relative humidity (3), in order to examine the influence of the DH content, temperature and moisture. Pseudo-polymorph stability was studied by preparing three samples each of A and MH
Figure 1. X-ray diffraction (XRD) patterns for the three-polymorphs of azithromycin. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
1338
BLANCO ET AL.
containing variable proportions of DH (0%, 2% and 7%, respectively). The six samples were stored in a stove at 608C in two moisture conditions: dry medium and saturation medium (100% RH) for 60 days. Apparatus and Software Near infrared spectra over the wavelength range 1100–2500 nm were recorded in the transflectance mode on a FOSS NIRSystems 6500 spectrophotometer equipped with a rapid content analyzer (RCA) module. The instrument was governed and data processed by using the software Vision v. 2.21 from FOSS NIRS ystems (Laurel, MD). Samples were placed in glass cuvettes and turned over with a spatula prior to recording triplicate spectra (averages of 32 scans) over the range 1100–2500 nm. The average of each triplicate spectrum was subjected to various treatments. The reference spectrum was obtained from an empty glass cuvette. Partial least-squares regression (PLS1) and principal component analysis (PCA) were performed by using the software Unscrambler v. 7.8, from Camo Process (Trondheim, Norway). Calibration models were constructed by cross-validation, using the leave-one-out method; the number of PLS components used was that minimizing the variation of the residual variance with the number of factors. The best model was taken to be that resulting in the lowest relative standard errors of calibration (RSEC) and prediction (RSEP).10 Moisture in the samples was determined on a 716 DMS Titrino Karl–Fisher titrator from Metrohm (Buckingham, UK). XRD patterns were obtained on an X’Pert Philips difractometer, using a Cu tube (l ¼ ˚ ) at 50 kV at 40 mA, a graphite secondary 1.5418 A monochromator, Bragg–Brentano geometry and a scintillation detector. XRD patterns were recorded over the 2y range 5–258, using a step width of 0.038 and an irradiation time of 3 s in each step.
Table 1. Figures of Merit of the Univariate Calibration by XRD (2y ¼ 13.0558) for Determination of Dihydrate (DH) in MH Samples (MH) of Azithromycin Concentration Range Calibration line Detection limit Quantification limit
0%–5% X ¼ 345.5 (13.8) þ 37.9 (4.9) C r ¼ 0.992 LD ¼ 3 sa/b ¼ 0.5% LQ ¼ 10 sa/b ¼ 1.5%
X, counts; C, concentration of DH (wt%); , confidence interval; sa, standard deviation of the intercept; b, slope.
were used to estimate the LD and LQ for these XRD methods. Those for DH in A were reported in a previous paper10 and those for DH in MH are given in Table 1.
Stability Study Figure 2 shows the NIR spectra for the three azithromycin pseudo-polymorphs. The greatest differences occur in the region from 1800 to 2200 nm, where the strongest band for water (viz. its combination band) appears. The anhydrous form exhibits a weak band corresponding to adsorbed water, whereas the crystalline forms (MH and DH) exhibit stronger water bands with one and two absorbance maxima, respectively, corresponding to crystallization and adsorbed water. The crystalline forms, MH and DH, possess a theoretical water content of 2.3% and 4.6% (wt %), respectively. The slight hygroscopicity of MH and A raises their water content to 1%–3% (wt %) without altering the pseudo-polymorph.
RESULTS AND DISCUSSION The three azithromycin pseudo-polymorphs were characterized from the XRD patterns of Figure 1. Samples of amorphous azithromycin (A) and the MH containing known amounts of the DH were used to construct a univariate calibration model for the DH content in the other two pseudopolymorphs. The regression lines thus obtained JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
Figure 2. SNV-corrected absorbance spectra for samples of the three pseudo-polymorphs of azithromycin.
POLYMORPHISM BY NIR
1339
Figure 3. (a) Evolution of NIR spectra of amorphous sample (0% DH) over time. (b) Evolution of NIR spectra of monohydrate (MH) sample (0% DH) over time.
In order to determine the stability of the pseudopolymorphs under different conditions, samples of the three azithromycin forms were stored in a stove at 608C prior to recording their NIR spectra, which revealed that adsorbed water was lost from A and MH, but not from DH. The combined effect of temperature (608C) and moisture (100% RH) resulted in no alteration of DH—which can thus be deemed the most stable form of the three. On the other hand, it caused A and MH to adsorb moisture. Figure 3 shows the NIR spectra in the region 1800–2200 nm for a sample of A and one of MH that were subjected to the above-described conditions for 60 days. As can be seen, both pseudopolymorphs present different behaviors. First the form A shows an increase of the water band, follow by an increase of the band ca. 1955 nm that corresponds to the DH form, therefore the form A structure absorbed water, that facilitate the conversion into DH. The MH only exhibited an increase in the first water band during the first 24 h, after which it leveled off; therefore, MH initially adsorbed an amount of moisture that remained constant thereafter, so it underwent no polymorphic transformation and was thus stable form under these conditions. The influence of the presence of DH (2% and 7%) in A and MH was studied in samples subjected to the same temperature and moisture conditions as the pure pseudo-polymorphs. The presence of DH in the MH had no effect on the stability of this pseudo-polymorph. On the other hand, the rate of the polymorphic transformation of A into DH increased with increasing DH content (A þ 7%
DH > A þ 2% DH > A), so the presence of DH in A accelerates its polymorphic transformation. Quantitative Analysis (PLS Models) The qualitative analysis exposed the behavior of A and MH under the conditions used in the stability study. In this study we apply the PLS1 calibration models to quantify the DH and water contents in samples of A and MH. However, in previous work, a method for determining DH and moisture contents of 0%–10% in samples of A was developed.10 The models examined in this work had to be adjusted to span the content range 0%–100%, however. The best results were obtained from second-derivative spectra in the region 1800– 2200 nm and three PLS factors. The DH and moisture contents in samples of MH were quantified by using PLS1 models involving different wavelength ranges (1100–2500 nm and 1800–2200 nm) spectral pretreatments (SNV, first and second derivative) and numbers of factors. The best results for both analytes (i.e., those exhibiting the lowest percent RSEC and RSEP values) are shown in Table 2. The previous PLS1 calibration models were used to determine DH and moisture in samples of A and MH. Figure 4 shows the predicted amounts of DH and water in samples of A containing 0%, 2%, or 7% DH. As can clearly bee seen, the contents in both moisture and DH increased with time, which confirms the occurrence of a polymorphic transformation. The DH contents of MH samples containing 0%, 2%, or 7% DH predicted by the proposed model JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
1340
BLANCO ET AL.
Table 2. Salient Features of the PLS Models Used in Determination of Dihydrate (DH) and H2O in Monohydrate Samples (MH) of Azithromycin Model Wavelength range (nm) Spectral pretreatment Concentration range PLS-factors Calibration
Prediction
DH
H2O
1100–2500 2nd derivative 0%–13.3% 4 n ¼ 39 CNIR ¼ 0.0691 þ 0.9863 CREF r ¼ 0.993 RSEC (%) ¼ 6.8 RMSEC (%) ¼ 0.4 n ¼ 44 CNIR ¼ 0.0584 þ 0.9885 CREF r ¼ 0.989 RSEP (%) ¼ 7.8 RMSEP (%) ¼ 0.4
1800–2200 2nd derivative 2.9%–5.4% 2 n ¼ 33 CNIR ¼ 0.0972 þ 0.9771 CREF r ¼ 0.988 RSEC (%) ¼ 2.5 RMSEC (%) ¼ 0.1 n ¼ 26 CNIR ¼ 0.1244 þ 1.0287 CREF r ¼ 0.977 RSEP (%) ¼ 2.7 RMSEP (%) ¼ 0.1
C, concentration (wt%).
Figure 4. (a) Prediction of DH contents in amporphous samples. (b) Prediction of moisture contents in amorphous samples.
remained constant with time, which confirms that this pseudo-polymorph is stable. As can be seen in Figure 5, the weight of the samples increased roughly by 1% (adsorbed water) within the first 24 h, after which it leveled off. Based on both the qualitative and quantitative results, MH is stable and A is transformed into DH under high temperature and moisture conditions.
Validation The proposed analytical method for determining the DH content in MH samples by NIRS was validated with a view to its use by the JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
Figure 5. Prediction of moisture in MH samples.
POLYMORPHISM BY NIR
1341
Table 3. Results of the Validation for the Proposed Method for the Quantitation of Dihydrate in Monohydrate Samples Parameter Linearity
Accuracy
Repeatability
Intermediate precision
Detection limit Quantification limit
Procedure
Results
CNIR ¼ a þ b CREF n¼8 Range ¼ 0%–10% Paired t-test of difference between NIR and reference value (n ¼ 8)
a ¼ 0.19 0.44 b ¼ 1.03 0.08 r ¼ 0.996 Residuals average ¼ 0.05 s ¼ 0.31 texp ¼ 0.459 ttab(a ¼ 0.05) ¼ 2.365 x1 ¼ 3.1 x2 ¼ 4.4 x3 ¼ 6.8
Three samples in three concentrations level analyzed by the same operator. Calculation of percent CV Sample analyzed 3 days by two different operators. ANOVA and calculation of percent CV LD ¼ 3.sa/b LQ ¼ 10.sa/b
x ¼ 4.4
CV1 ¼ 2.7 CV2 ¼ 4.4 CV3 ¼ 3.5 CVtotal ¼ 4.0
Not differences between days and operators
CV ¼ 4.8 LD ¼ 0.2% LQ ¼ 0.8%
C, concentration of DH (wt%); , confidence interval; sa, standard deviation of the intercept; b, slope.
pharmaceutical industry in accordance with the International Conference of Harmonization (ICH) and the EMEA guidelines. This purpose must be considered in determining the validation parameters required (viz. selectivity, linearity, range, accuracy, precision—as repeatability, and intermediate precision—, and LD and LQ).
the characteristics from the validation procedure for each one parameter and they results. The LD and LQ are equivalent as the obtained by XRD and testify to the suitability of the method for the intended purpose.
CONCLUSIONS Selectivity We addressed the identification of MH and DH azithromycin. We constructed a library from spectra for pure MH and DH azithromycin from different production batches (n ¼ 48). We used the 2nd derivative spectral mode, the wavelength range of 1100–2500 nm and the correlation coefficients in wavelength space as discriminating parameter, with a threshold of 0.90. We assessed the selectivity of the identification library by using it to identify external samples of MH and DH azithromycin. All samples were correct identified.
Quantitative Parameters The results for the validation of quantitative parameters are shown in Table 3, that indicate
The use of the NIRS technique to study the accelerated transformation of azithromycin pseudo-polymorphs under accelerated temperature and moisture conditions (at 608C and 100% RH) revealed that the two crystalline forms (viz. the MH and DH) are stable, whereas the amorphous form evolves to the DH, which is the most stable of the three pseudo-polymorphs. The proposed method for quantifying the DH in MH samples testifies to the suitability of NIRS methodology for characterizing and determining azithromycin pseudo-polymorphs.
ACKNOWLEDGMENTS The authors are grateful to Spain’s Ministry of Science and Technology (MCyT) for funding JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
1342
BLANCO ET AL.
this research within the framework or Project BQU2003-04247.
REFERENCES 1. Grant DJW. 1999. Theory and origin of polymorphism. In: Brittain HG, editor. Polymorphism in pharmaceutical solids, edn. New York: Marcel Dekker, Inc. pp 1–10. 2. Bakeev KA. 2003. Near-infrared spectroscopy as a process analytical tool. Pharm Technol Europe 15:27–28, 30–32. 3. Note for guidance on the use of near infrared spectroscopy by the pharmaceutical industry. 2003. http://www.emea.eu.int/pdfs/human/qwp/ 330901en.pdf 4. ICH quality guidelines. Q2A, text on validation of analytical procedures 1994; Q2B, validation of analytical procedures: methodology. 1996. http:// www.ich.org 5. Davis TD, Peck GE, Stowell JG, Morris KR, Byrn SR. 2004. Modeling and monitoring of polymorphic transformations during the drying phase of wet granulation. Pharm Res 21:860–866.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 6, JUNE 2005
6. Rasanen E, Rantanen J, Jorgensen A, Karjalainen M, Paakkari T, Yliruusi J. 2001. Novel identification of pseudopolymorphic changes of theophylline during wet granulation using near infrared spectroscopy. J Pharm Sci 90:389–396. 7. Airaksinen S, Karjalainen M, Raesaenen E, Rantanen J, Yliruusi J. 2004. Comparison of the effects of two drying methods on polymorphism of theophylline. Int J Pharm 276:129–141. 8. Norris T, Aldridge PK, Sekulic SS. 1997. Determination of end-points for polymorph conversions of crystalline organic compounds using online nearinfrared spectroscopy. Analyst (Cambridge, United Kingdom) 122:549–552. 9. De Braekeleer K, De Maesschalck R, Hailey PA, Sharp DCA, Massart DL. 1999. Online application of the orthogonal projection approach (OPA) and the soft independent modeling of class analogy approach (SIMCA) for the detection of the end point of a polymorph conversion reaction by near infrared spectroscopy (NIR). Chemometrics and Intelligent Laboratory Systems 46:103–116. 10. Blanco M, Valde´s D, Bayod MS, Ferna´ndez-Mari F, Llorente I. 2004. Characterization and analysis of polymorphs by near-infrared spectrometry. Analytica Chimica Acta 502:221–227.