Anal. Chem. 2009, 81, 1404–1410
Method for the Determination of Pd-Catalyst Residues in Active Pharmaceutical Ingredients by Means of High-Energy Polarized-Beam Energy Dispersive X-Ray Fluorescence E. Marguı´,*,† K. Van Meel,‡ R. Van Grieken,‡ A. Buendı´a,§ C. Fonta`s,| M. Hidalgo,| and I. Queralt† Laboratory of X-Ray Analytical Applications, Institute of Earth Sciences “Jaume Almera”, CSIC, Sole´ Sabarı´s s/n, 08028 Barcelona, Spain, Department of Chemistry, University of Antwerp, Universiteitsplein 1, 2610 Antwerp, Belgium, Department of Quality Control, Medichem, S.A. Pol. Ind. Celra`, 17460 Celra`, Spain, and Department of Chemistry, University of Girona, Campus Montilivi, 17071 Girona, Spain In medicinal chemistry, Pd is perhaps the most-widely utilized precious metal, as catalyst in reactions which represent key transformations toward the synthesis of new active pharmaceutical ingredients (APIs). The disadvantage of this metal-catalyzed chemistry is that expensive and toxic metal residues are invariably left bound to the desired product. Thus, stringent regulatory guidelines exist for the amount of residual Pd that a drug candidate is allowed to contain. In this work, a rapid and simple method for the determination of Pd in API samples by high-energy polarized-beam energy dispersive X-ray fluorescence spectrometry has been developed and validated according to the specification limits of current legislation (10 mg kg-1 Pd) and the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH guidelines). Sample and calibration standards preparation includes a first step of homogenization and then, in a second step, the pressing of the powdered material into pellets without any chemical treatment. The use of several synthetic calibration standards made of cellulose to simulate the API matrix appears to be an effective means to obtain reliable calibration curves with a good spread of data points over the working range. With the use of the best measuring conditions, the limit of detection (0.11 mg kg-1 Pd) as well as the limit of quantitation (0.37 mg kg-1 Pd) achieved meet rigorous requirements. The repeatability of the XRF measurement appeared to be less than 2%, while the precision of the whole method was around 7%. Trueness was evaluated by analyzing spiked API samples at the level of the specification limit and calculating the recovery factor, which was better than 95%. To study the applicability of the developed methodology for the intended purpose, three batches of the studied API were analyzed for their Pd content, and the attained results were comparable to those obtained by the daily * Corresponding author. E-mail:
[email protected]. Fax: (+34)-934110012. † Institute of Earth Sciences “Jaume Almera”, CSIC. ‡ University of Antwerp. § Medichem. | University of Girona.
1404
Analytical Chemistry, Vol. 81, No. 4, February 15, 2009
routine method (acid digestion plus atomic spectroscopy) used in most pharmaceutical laboratories. The essential characteristics for pharmaceuticals are safety and efficacy. Consequently, the determination of potential impurities in different stages of the manufacturing processes, and especially in the final product, is crucial.1 The monitoring of heavy metals which generally originate from various sources and phases (i.e., raw materials, reagents, solvents, electrodes, catalysts, reaction vessels, plumbing, and other equipment used during the synthesis of pharmaceuticals) is an important activity in pharmaceutical industry, not only because of the ability of heavy metals to catalyze decomposition but also their potential for toxicity.2 The use of Pd-derived catalysts in the synthesis of fine chemicals, pharmaceutical intermediates, and active pharmaceutical ingredients (APIs) has become quite common in the last few decades. The number of Pd-catalyzed synthetic reactions available to chemists has provided access to more complex structures in fewer steps and with less waste, due to the catalytic nature of many of the methods. However, an unfortunate side effect of using Pd is its potential to remain in the desired compound after isolation.3 For API specimens, there are strict guidelines to limit the levels of heavy metals, including Pd, in the drug substance. At the beginnings 2008, the European Agency for the Evaluation of Medicines (EMEA) presented a guideline on the specification limits for residues of metal catalysts or metal reagents that has come into effect in September 2008.4 This guideline classifies metals in three categories based on their individual levels of safety concern, and concentration limits are set according to the maximal daily dose, duration of treatment, route of administration, and permitted daily exposure (see Table 1). As it is shown, Pd is included in the class 1 group which comprises metals of significant safety concern. The limit for oral administration of this metal has been established at 10 mg kg-1 in the API. (1) Hulse, W. L.; Grimsey, I. M.; De Matas, M. Int. J. Pharm. 2008, 349, 61– 65. (2) Nageswara Rao, R.; Kumar Talluri, M. V. N. J. Pharm. Biomed. Anal. 2007, 43, 1–13. (3) Garrett, C. E.; Prasad, K. Adv. Synth. Catal. 2004, 346, 889–900. (4) Committee for medicinal products for human use (CHMP). Guideline on the Specification Limits for Residues of Metal Catalysts or Metal Reagents; Doc. No. EMEA/CHMP/SWP/4446/2000); European Medicines Agency: London, U.K., February 2008. 10.1021/ac8021373 CCC: $40.75 2009 American Chemical Society Published on Web 01/22/2009
Table 1. Class Exposure and Concentration Limits for Individual Metal Catalysts and Metal Reagents According to the European Agency for the Evaluation of Medicines (EMEA, September 2008) oral exposure
parental exposure
PDE concentration PDE concentration classification (µg day-1)a (mg kg-1) (µg day-1)a (mg kg-1) class 1A: Pt, Pd class 1B: Ir, Rh, Ru, Os class 1C: Mo, Ni, Cr, Vb class 2: Cu, Mn class 3: Fe, Zn
100 100
10 10
10 10
1 1
250
25
25
2.5
2500 1300
250 1300
250 1300
25 130
a PDE, permitted daily exposure. b Subclass limit, the total amount of listed metals should not exceed the indicated limit.
For the determination of each of the specified metals, an appropriate and validated method should be used in relation to the limit to be applied. All pharmacopoeias include a test for heavy metals, which is commonly carried out by sulfide precipitation in a weakly acidic medium and visual comparison of the color of a simultaneously and similarly treated standard solution of lead.5,6 Serious limitations of this visual semiquantitative test are the lack of sensitivity, specificity, and recovery to monitor properly the actual levels for some metals. Recent papers have reported recoveries for elements such as As, Cd, Mo, Pd, Pt, and In between 30% and 50%.7 For this reason, spectroscopic techniques including electrothermal atomic absorption spectrometry (ETAAS),8 inductively coupled plasma atomic emission spectrometry (ICP-AES),9 and inductively coupled plasma mass spectrometry (ICPMS)10 are preferred in some applications. However, this type of instrument is basically designed for the analysis of liquid samples. Therefore, solid pharmaceutical samples have to be brought into solution in order to satisfy the need of introduction systems of most spectroscopic techniques used. Most relevant sample preparation methods used for the determination of Pd in APIs by atomic spectrometry are summarized in Table 2. Sample preparation is usually a critical step in pharmaceutical samples because of the complex and sometimes harmful sample matrixes. Frequently, pharmaceutical chemicals are not sufficiently soluble in deionized water or dilute acids and solubilization using organic chemicals7,11,12 or concentrated acids8,10 is required. When the sample is difficult to dissolve, a digestion method using different acid mixtures should also be considered for the analysis of metal impurities.9 Although very effective, the digestion approach is time-consuming and requires the use of corrosive and harmful reagents. A promising alternative could be the use of solid (5) The United States Pharmacopoeia, USP30. Chemical Tests. Limit Tests. Heavy Metals; 2007, pp 146-147. (6) European Pharmacopoeia 5.0. Limit Test. Heavy Metals (Method 2.4.8); 2005, pp 104-107. (7) Lewen, N.; Mathew, S.; Schenkenberger, M.; Raglione, T. J. Pharm. Biomed. Anal. 2004, 35, 739–752. (8) Niemela¨, M.; Kola, H.; Eilola, K.; Pera¨ma¨ki, P. J. Pharm. Biomed. Anal. 2004, 35, 433–439. (9) La´sztity, A.; Kelko´-Le´vai, A´; Varga, I.; Zih-Pere´nyi, K.; Bertalan, E´. Microchem. J. 2002, 73, 59–63. (10) Lewne, N.; Schenkenberger, M.; Larkin, T.; Conder, S.; Brittain, H. G. J. Pharm. Biomed. Anal. 1995, 35, 879–883. (11) Jia, X.; Wang, T.; Wu, J. Talanta 2001, 54, 741–751. (12) Wang, T.; Walden, S.; Egan, R. J. Pharm. Biomed. Anal. 1997, 15, 593– 599.
state techniques such as X-ray fluorescence spectrometry (XRF). The fact that the analysis could be performed directly on the solid sample entails less sample manipulation and lower amounts of reagents and therefore, cost and time savings for the manufacturers. XRF has been a popular technique for major elemental analysis in geological samples to avoid complicated acid-digestion procedures. In particular, the speed, accuracy, and versatility of XRF are the most important features among the many that have made it a very mature analytical tool in this field.13 However, it is widely recognized that conventional XRF instrumentation, at lower excitation energies, does not allow the excitation of Pd for K-line emission. Even with sufficiently high power tubes, only the bremsstrahlung will be capable of exciting the K-lines of high atomic number elements, resulting in poor sensitivity. Analyzing those elements by their L-lines give rise also to a limited sensitivity (due to the lower fluorescent yield) and more complicated spectra, with a higher probability of spectral overlap. This results in higher detection limits and uncertainty. In the present work, the use of new equipment based on highenergy polarized-beam energy-dispersive X-ray fluorescence (HEP-EDXRF) is proposed. In this instrument, the combined use of high voltage Gd X-ray tube and a high-energy Ge semiconductor detector allows performing EDXRF analysis using K-lines of high atomic number elements such as Pd and, thus, the improvement of both selectivity and limits of detection. Moreover, in the instrument used, the primary beam is scattered by a secondary target. The use of suitable secondary targets in a three-dimensional polarizing geometry for excitation not only allows reducing the intensity of the measured continuum radiation (by strongly reducing the scattered tube radiation), thus significantly increasing the signal-to-noise ratio but also increases the X-ray production due to the choice of quasi monochromatic radiation with energy close to the photoelectric absorption edge of a given element of interest. Therefore, compared to other XRF techniques, it has the advantages of increased sensitivity and specificity for trace heavy metal analysis as has been demonstrated in previous works.14-17 The main goal of the present research was to develop a fast method with as simple sample preparation as possible, without losing analytical performance compared to atomic spectroscopic techniques commonly applied for metal residues determination at the pharmaceutical laboratories. The active pharmaceutical ingredient analyzed in this project was a new triazole antifungal medication that is generally used to treat serious, invasive fungal infections. Pd is used as catalyst in the synthesis of this drug and is, therefore, one of the potential impurities in the final product that has to be routinely monitored. To facilitate the adoption of new or improved analytical methodologies which update or replace current practices, the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) includes a document on the characteristics for consideration (13) Hettipathirana, T. D. X-Ray Spectrom. 2001, 30, 330–337. (14) Spolnik, Z.; Belikov, K.; Van Meel, K.; Adriaenssens, E.; De Roeck, F.; Van Grieken, R. Appl. Spectrosc. 2005, 59, 1465–1469. (15) Van Meel, K.; Smekens, A.; Behets, M.; Kazandjian, P.; Van Grieken, R. Anal. Chem. 2007, 79, 6383–6389. (16) Van Meel, K.; Fonta`s, C.; Van Grieken, R.; Queralt, I.; Hidalgo, M.; Marguı´, E. J. Anal. At. Spectrom. 2008, 23, 1034–1037. (17) Marguı´, E.; Fonta`s, C.; Van Meel, K.; Van Grieken, R.; Queralt, I.; Hidalgo, M. Anal. Chem. 2008, 80, 2357–2364.
Analytical Chemistry, Vol. 81, No. 4, February 15, 2009
1405
Table 2. Analytical Procedures for Pd Determination in Pharmaceutical Compounds Using Atomic Spectroscopic Techniques sample preparation
pharmaceutical compound
procedure
medium
technique
LOD (mg kg-1)
ref
API enalapril maleate fosinopril sodium methotrexate (MTX) bulk pharmaceutical chemicals bulk pharmaceutical chemicals
dissolution dissolution dissolution microwave digestion dissolution dissolution
2-butoxy ethanol/water (25:75 v/v) 1:1 HNO3, 0.3M 2-butoxy ethanol/water (25:75 v/v) H2SO4 + HNO3 acetonitrile, dimethylsulfoxide 70% HNO3
ICPMS ICPMS ICPMS DCP-OESa ETAAS ETAAS
0.18 0.015 0.1 0.3 not reported 0.7, 3.5
7 8 9 10 11 12
a
DCP-OES, direct current plasma-optical emission spectrometry.
during the validation of any analytical procedure used in the analysis of impurities in pharmaceuticals for human use.18 Typical validation characteristics which should be considered according to this guideline (specificity, detection limit, quantification limit, linearity, trueness and precision) were also evaluated in this study to test the real applicability of the developed methodology for the intended purpose. EXPERIMENTAL SECTION Sample and Calibration Standards Preparation for XRF Analysis. As already stated above, the target active pharmaceutical ingredient was a triazole antifungal drug substance that is presented in the form of a white to light-colored powder. Since this pharmaceutical compound is, at present, under development, its chemical structure is proprietary and cannot be displayed. To the best of our knowledge, there were no suitable pharmaceutical reference materials available for Pd determination. Because of, the utilization of calibration standards prepared in the laboratory with commercially available pure elements or compounds was studied as an alternative for calibration purposes. Taking into account that the X-ray interactions depend on the effective atomic number (Zeff) of the matrix, this term was used to select the best candidate compound to simulate the API matrix. Zeff was calculated using the following expression:19
∑w Z A
k k k
Zeff )
k
∑w A
(1)
k k
k
where wk is the weight fraction, Ak is the atomic mass, and Zk is the atomic number of element k. It was found that the effective atomic number for the API (Zeff ) 6.58) was really similar to that calculated for cellulose (Zeff ) 6.68). Therefore, high-purity microcrystalline cellulose (cellulose powder, Sigma-Aldrich, Buchs SG, Schweiz) was employed as the synthetic matrix. A stock cellulose standard containing 50 mg kg-1 of Pd was prepared by weighting 25 g of cellulose and adding 25 mL of a Pd solution at 50 mg L-1 (prepared from a commercially available stock solution of Pd at 1016 mg L-1 purchased from (18) International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH). Validation of Analytical Procedures: Text and Methodology Q2(R1); November 2005. (19) Padilla, R.; Van Espen, P.; Quintana, A. A. X-Ray Spectrom. 2004, 33, 74– 82.
1406
Analytical Chemistry, Vol. 81, No. 4, February 15, 2009
Aldrich Chemical Co., Inc. Milwaukee, WI). Dilutions using appropriate amounts of cellulose were carried out to obtain calibration standards over the studied working range. After drying at room temperature, the spiked cellulose was homogenized in an agate mortar. Then, the calibration standards were stored in polypropylene containers until analysis. For trueness studies, API samples were prepared by following the same described spiking procedure. For XRF analysis, samples and calibration standards were prepared in the form of pressed powder pellets. In a first step, and taking into account the crystallinity of the powder, a grinding process using an automatic mixer/mill (Mixer/mill MM 301, Retsch, Aartselaar, Belgium) with polystyrene balls was carried out. Considering the morphology of the studied matrixes, the preparation of the pellets for the API samples was performed without the addition of a binder. To assess the type of attenuation for Pd-KR, pellets of different thickness were prepared from a stock cellulose standard containing 50 mg kg-1 of Pd. The obtained results evidence that the condition of an “infinite thick” sample can be reached for pellet thickness of about 7 mm. However, since the amount of pharmaceutical material was scarce, we chose to work with thinner samples (around 5 mm). To avoid errors in quantitative measurements due to the condition of “intermediate thickness”, pellets (diameter 10 mm) were prepared weighing a constant mass of sample (1 g) and pressed at a constant pressure (10 tons) using a manual hydraulic press. All accessories were delivered by Specac (Kent, United Kingdom). Instrumentation. For the HE-PEDXRF analysis, the Epsilon-5 instrument (PANalytical, Almelo, The Netherlands) was used. Epsilon-5 is equipped with a 600 W Gd-anode with a voltage ranging from 25 to 100 kV and a current from 0.5 to 24 mA. There are 13 secondary targets (Al, CaF2, Ti, Fe, Co, Ge, KBr, Zr, Mo, Ag, CsI, CeO2, and W) and two Barkla-scatterers (Al2O3 and B4C). It is possible to use a primary beam filter between tube and target; the following filter materials are present: Al 100 µm, Al 500 µm, Cu 250 µm, Zr 125 µm, and Mo 250 µm. Detection is performed with a high purity Ge-detector with an energy range from 0.7 to 200 keV and a resolution 95%) with recovery values of 97% when using Al2O3 (9.70 ± 0.30 mg kg-1 Pd) and 96% using the CsI as secondary target (9.62 ± 0.62 mg kg-1 Pd). Thus, the trueness of the procedure and the absence of matrix effects were confirmed for this kind of pharmaceutical sample. Precision. Precision studies were also carried out by analyzing spiked API samples at the level of 10 mg kg-1. For repeatability conditions, the same pellet was successively measured six times (Table 4a). This uncertainty is related to the instrument and counting statistics and it was found that, for both targets, relative standard deviations (RSD) were below 2%. In order to study the overall precision of the proposed method, six independent pellets of the spiked sample were prepared and measured under the same experimental conditions. Calculated RSD results are given in Table 1408
Analytical Chemistry, Vol. 81, No. 4, February 15, 2009
4b. A difference in precision for the two targets studied was observed. It appeared that the precision of the Al2O3 was about 3.0%, while the precision for the CsI target was 6.5%. It could be stated that the experimental procedure is of good quality keeping in mind that the values obtained following this procedure take into account the uncertainty due to sample preparation, instrument, and counting statistics. Moreover, from data obtained and taking into account the variance models (σtotal2 ) σ12 + σ22 + σ23 +...+ σN2), the influence of the sample preparation on the uncertainty was estimated using the following expression: σtotal2 ) σsample preparation2 + σXRF measurement2
(2)
It can be assumed that the total variance is given by the results of the six pellets, since it is subjected to both the influence of the sample preparation (six different pellets) and to the influence of the XRF measurements (six separate measurements). On the other hand, the results of one particular pellet rule out the influence of the sample preparation. With the use of the values for the RSD shown in Table 4, it can be calculated that the influence of the sample preparation is lower than 6.5%.
Figure 3. Calibration curves and residuals graphs for Pd determination using Al2O3 and CsI targets (Each point represents the average of the three pellets measurements). Table 3. Calibration Parameters for Consideration during Linearity Evaluation (ICH Guideline) target
range (mg kg-1, Pd)
regression coefficient
Al2O3
0.5-20
0.997
2-20
0.995
CsI
a RSS, residual sum of squares; a, intercept; b, slope. calculated (triplicate analysis).
a ± Sa
b
b ± Sb
RSS (Σei2)a
0.39 ± 0.16
0.618 ± 0.016
0.85
-1.50 ± 0.84
1.961 ± 0.076
(b) 6 pellets/1 measurement
Al2O3 target
CsI target
Al2O3 target
CsI target
1 2 3 4 5 6
6.72 6.40 6.45 6.46 6.54 6.61
18.3 18.4 18.5 18.4 18.6 18.4
6.50 6.34 6.11 6.35 6.45 6.68
20.4 18.4 17.5 18.3 18.3 20.5
average std deviation RSD (%)
6.53 0.12 1.8
18.4 0.11 0.58
6.41 0.19 3.0
18.9 1.23 6.5
replicates
a
7.1 (0.5) 5.4 (10) 10.2 (12) 12.3 (2) 4.3 (10) 8.5 (12)
10.9
Values in parentheses are the Pd concentrations (mg kg-1) for which the RSD was
Table 4. Precision Study for Pd Determination Measuring a Spiked API Sample at the Level of 10 mg kg-1 of Pda (a) 1 pellet/6 measurements
RSD in %b
The values are expressed as Pd XRF-response (Cps mA-1).
Reproducibility of the HE-P-EDXRF method over time was evaluated by measuring the same specimen (previously used in the repeatability study) over a period of 1 month (n ) 9). The calculated RSD values were for both targets ∼6%.The very limited differences in the fluorescent intensities after several irradiations demonstrated the absence of damage of the pellets. The XRF spectrometer used in the present study exhibits very low sample heating due to the combination of low power (600 W) and threedimensional optics. As a result, delicate organic samples can be reanalyzed leaving the sample unaffected.
Table 5. Comparison of Mean Pd Concentrations in Three Batches of the Target API Analyzed by the Proposed Method and by the Daily Routine Method (Acid Digestion Plus Inductively Coupled Plasma Atomic Emission Spectrometry) Pd concentration (mg kg-1)a this method sample
Al2O3
CsI
reference method
API-1 API-2 API-3
0.35 ± 0.06 0.86 ± 0.07 0.30 ± 0.02
