The comparative verification of calibration curve and background fundamental parameter methods for impurity analysis in drug materials

Transcription

1 Special issue article Received: 1 September 2016 Revised: 26 May 2017 Accepted: 26 May 2017 Published online in Wiley Online Library: 29 June 2017 (wileyonlinelibrary.com) DOI /xrs.2788 The comparative verification of calibration curve and background fundamental parameter methods for impurity analysis in drug materials Hiroaki Furukawa, a * Naoto Ichimaru, b Keijiro Suzuki, a Makoto Nishino, a Jennifer Broughton, c Johan Leinders d and Hirotomo Ochi a The United States Pharmacopeia (USP) has established general analysis guidelines of X-Ray fluorescence (XRF) methodology in chapter <735>, XRF spectrometry. Meanwhile, the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH Q3D) guideline, presenting a policy for limiting 24 elements such as Cd, Pb, and As in drug products and pharmaceutical ingredients, has now reached Step 5 (Implementation). The limit values of ICH Q3D are strict, for example, near or less than 1 μg/g for Cd and Pb. Moreover, there are other difficulties for accurate analysis, such as different types of matrices and the very small sample volumes in the synthesis research process for new product development in the pharmaceutical industries. Energy dispersive XRF can correct the influence of various matrices and volumes and also has the advantages of nonchemical pretreatment and nondestructive analysis. Advancements in the quantitation by fundamental parameters (FP) method using theoretical calculations, background (BG)-FP method, leads to reduction in the number of standard samples and effectiveness on the practical elemental impurity control in pharmaceutical researches and inspections. In order to show the applicability of using the BG-FP method, in this study, the quantitative values of some hazardous inorganic impurities for Cd, Pb, As, V, Co, and Ni in a cellulose matrix were compared by calibration curve method and BG-FP method through the verification of USP <735>. Both methods can meet the acceptance criteria for accuracy and precision in USP <735>. Copyright 2017 John Wiley & Sons, Ltd. 382 Introduction The United States Pharmacopeia (USP) has been improved global health through public standards that help ensure the quality, safety, and benefit of drugs and foods. USP has established general analysis guidelines of X-ray fluorescence (XRF), that is, XRF methodology in USP <735>. [1] USP <232>, Elemental Impurities Limits, and USP <233>, Elemental Impurities Procedures, for impurity control in drug substances and excipients will be enacted in January The sensitivity levels of energy dispersive XRF (EDX) for some elements are close to meeting the specification of USP <232>, and in the future, EDX is expected to be used for impurity control analysis in the pharmaceutical industry as an alternative method to USP <233>. USP <232> defines the default values of permitted daily exposure (PDE) for oral drug products of Class 1 (Cd, Pb, As, and Hg) and Class 2A (V, Co, and Ni) assessment elements based on the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH Q3D). [2] It is necessary to define the maximum daily intake to convert to concentration limits because the unit of PDE is μg/day. ICH Q3D guideline advocates some conversion methods such as Option 1 or Option 2a. [3] Option 1 is for common permitted concentration limits of elements across drug product components for drug products with daily intakes of not more than 10 g. Option 2a is for common permitted concentration limits across drug product components for a drug product with a specified daily intake. The most important thing is to change the concentration limits depending on the conversion method. Table 1 shows the PDE and concentration limits by Options 1 and 2a for the above assessment elements. The maximum daily intake assumes 1 g for Option 2a. USP <735> has been enacted in May The contents consist of operational qualification and performance qualification. The specification of operational qualification checks basic instrument parameters such as peak position, detector resolution, and count rate using specific metal samples. Performance qualification also has some verification items; however, this study focuses on the practical items as follows [1] : 1. Linearity Analysts should demonstrate a linear relationship between the analyte concentration and corrected XRF response by preparing no fewer than five standards at concentrations that encompass the anticipated concentration of the test * Correspondence: Hiroaki Furukawa, Shimadzu Corportion 1 Nishinokyo Kuwabara-cho, Nakagyo-ku, Kyoto , Japan. furukawa@shimadzu.co.jp Presented at EXRS2016 European Conference on X ray Spectrometry, Gothenburg, Sweden, June a Shimadzu Corporation, 1 Nishinokyo Kuwabara-cho, Nakagyo-ku, Kyoto , Japan b Shimadzu Techno-Research Inc., 1 Nishinokyo Shimoai-cho, Nakagyo-ku, Kyoto , Japan c Shimadzu Research Laboratory (Europe) Ltd., Wharfside, Trafford Wharf Road, Manchester, UK d Shimadzu Europa GmbH, 6-10 Albert-Hahn-Str, 47269, Duisburg, Germany X-Ray Spectrom. 2017, 46, Copyright 2017 John Wiley & Sons, Ltd.

2 Impurity analysis in drug materials Table 1. Permitted daily exposure (PDE) and concentration limits by Options 1 and 2a for Class 1 and 2A elements [2] Element PDE for oral drug products (μg/day) Concentration limits (μg/g) by Option 1 (maximum daily intake is 10 g/day) Concentration limits (μg/g) by Option 2a (maximum daily intake is 1 g/day) Cadmium, Cd Lead, Pb Arsenic, As Mercury, Hg Vanadium, V Cobalt, Co Nickel, Ni sample. The standard curve then should be evaluated using appropriate statistical methods such as least squares regression. The correlation coefficient (R), y-intercept, and slope of the regression line must be determined. Acceptance criteria: R is not less than Accuracy/specificity Analysts can determine accuracy by conducting recovery studies using the appropriate matrix spiked with known concentrations of elements. It is also an acceptable practice to compare assay results obtained using the XRF method under validation to those from an established analytical method. Acceptance criteria: 70.0% to 150.0% recovery 3. Quantitation limit The limit of quantitation can be estimated by calculating the standard deviation of not less than six replicate measurements of a blank and multiplying by 10. Acceptance criteria: The analytical procedure should be capable of determining the analyte precisely and accurately at a level equivalent to 50% of the specification. Table 2. Calibration sample and verification sample information Sample type Additive elements Concentration (μg/g) Calibration sample (water solution) Verification sample (cellulose) Pb, Cd, As, V, Co, and Ni Note. BG-FP = background fundamental parameters. 0 (blank), 1, 3, 5, and 10 for calibration curve method 100 for BG-FP method 5and10 4. Robustness The reliability of an analytical measurement should be demonstrated by deliberate changes to experimental parameters. Acceptance criteria: The measurement of a standard or sample response following a change in experimental parameters should differ from the same standard measured using established parameters by not more than ±20%. Experimental EDX spectroscopy measurements were carried out using the Shimadzu EDX-7000 bench top spectrometer (Shimadzu Co., Kyoto, Japan). [4] The spectrometer is equipped with a rhodium tube (1,000 μa, 50 kv, and 50 W) and a silicon drift detector with an active area of 25 mm 2, 0.5 mm thickness. The energy resolution at the Mn Kα line is 135 ev. The calibration and samples were measured using a counting time of 300 and 1,200s for quantitation limit test and an excitation X-ray beam size of 10 mm diameter. The calibration samples were generated from a water solution spiked at seven concentrations, including zero concentration (pure water), of Cd, Pb, As, V, Co, and Ni using standard solutions for atomic absorption spectroscopy. The verification samples were generated from cellulose powder spiked at two concentrations of the above six elements using standard solutions for atomic absorption spectroscopy. The standard solutions were added to the blank cellulose samples and mixed in an agate mortar to prevent adhesion to the walls. In order to check for homogeneity, a small amount of each spiked cellulose sample was taken and divided into three subsamples, which were measured for the quantitative amounts of target elements present. The three subsamples were then mixed together before being redivided into three new subsamples for measurement. The mixing, division into three, and measurement were then repeated (nine measurements in total). If the quantitative values of target elements were consistent for the different subsamples, then the sample was regarded as homogeneous. The prepared samples and the measurement conditions are summarized in Tables 2 and 3. The measurement conditions were used for the above analyses. The 2.0 g cellulose powder samples (Figure 1) were placed directly into a disposable cup (32 mm diameter) with a 5 μm polypropylene support film. The samples were prepared using a tapping method to settle them in the cups in an effort to reduce voids in the sample packing. The 0.1 g cellulose powder samples were also placed directly into disposable cups with a 5 μm polypropylene support film for robustness testing. The 0, 1, 3, 5, and 10 μg/g samples were used to generate a calibration curve Table 3. Measurement conditions Elements Voltage (kv) Automatic current adjustment Measurement time (s) Filter Scattering X-rays Element line Cd 50 Yes 300 Cu 25 kev Cd Kα Pb As 1,200 for quantitation limit Ag 15 kev PbLβ1 As Kα Co Ni 10 kev Co Kα Ni Kα V None 5 kev V Kα Water cellulose Rh Compton Rh Rayleigh main component 383 X-Ray Spectrom. 2017, 46, Copyright 2017 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/xrs

3 H. FURUKAWA ET AL. Figure 1. Cellulose sample preparation for each of the above elements, and the 100 μg/g sample was used to determine the fundamental parameters (FP) coefficient for the background (BG)-FP method. Continuous scattered X-rays at 5, 10, 15, and 25 kev were used as an internal standard to correct for sample thickness effects. The quantitation of impurity elements was calculated by the ratio of fluorescent X-rays for each element with continuous scattered X-rays. The water matrix component was programmed as the chemical formula H 2 O. The quantitation of the water matrix was calculated by the FP coefficient parameter, which is based on the correlation between theoretical and measured intensities using the theoretical and measured values of the ratio of rhodium Compton and Rayleigh scattering X-rays, because this ratio has a linear relationship to the average atomic number (Figure 2). This can be used to correct for matrix effect differences between water and cellulose using the ratio of rhodium Compton and Rayleigh scattering X-rays. The intensities of fluorescent X-ray peaks were calculated by a curve fitting method, and the intensity of Compton and Rayleigh scattering X-ray peaks were calculated by a region of interest method. Table 1 includes mercury; however, mercury was not evaluated in the above validation because it evaporates too easily during sample preparation. BG-FP method Figure 2. Relationship between the average atomic number (mean Z) and the intensity ratio of rhodium Compton and Rayleigh scattering X-rays (C/R ratio). Data measured with Shimadzu EDX Gray shaded band shows 95% confidence interval of the linear regression The FP method is a very useful quantitative technique. [5] Akey advantage of using this method is that the number of standard samples used is reduced compared to the calibration curve method. [6] The FP method uses calculations that require all components in the sample to be known. In an organic sample, the organic matrix generally includes hydrogen, carbon, and oxygen, which cannot easily be detected by conventional EDX systems. In XRF, the determined quantitative value of trace elements is very sensitive to factors like changing the quantitative value of the organic matrix, as well as sample shape and thickness. To reduce the influence of these factors, the BG-FP method was used [7]. [8] This method uses not only the fluorescent X-rays of the target elements but also the scattered X-rays of continuous or characteristic X-rays, which come from the X-ray source. Equation (1) shows that the theoretical intensities of scattered X-rays can be calculated using the FP coefficient parameter between measured and theoretical intensities, as well as the calculation of the theoretical intensities of fluorescent X-rays. The theoretical intensities of scattered X-rays (T) are expressed by summation of theoretical Rayleigh (R) and Compton (C) scattering X-rays of both trace elements and organic matrix in the sample. 384 Figure 3. Typical spectrum of 100 μg/g lead contained in cellulose sample wileyonlinelibrary.com/journal/xrs Copyright 2017 John Wiley & Sons, Ltd. X-Ray Spectrom. 2017, 46,

4 Impurity analysis in drug materials Hence, these scattered X-rays can be expressed as a function of concentration of elements in the sample [9],[10] : T ¼ C þ R; (1) where the theoretical intensities of Rayleigh and Compton scattering X-rays, R and C, can be given as follows: R ¼ 1 sinψ C ¼ 1 sinψ I λ R i i μ λ sin Φ þ μ λ sin Ψ I 0 λ C i i μ 0 λ sin Φ þ μ λ sin Ψ ; (2) ; (3) where R, C = theoretical intensity of Rayleigh and Compton scattering X-rays (count); λ, λ 0 = wavelength of X-rays (Å); 0 I λ,i λ = theoretical intensity of incident X-rays of wavelength λ, λ 0 (count); R i = efficiency coefficient of Rayleigh scattering from element i (cm 2 /g); C i = efficiency coefficient of Compton scattering from element i (cm 2 /g); μ λ, μ λ 0 = mass absorption coefficient of the sample for wavelength λ, λ 0 (cm 2 /g); Φ = angle of incident X-ray (rad); Ψ = detection angle of scattered X-ray (rad). The theoretical intensity of scattered X-rays at any energy position can be calculated using the FP coefficient parameter, because it is well known that this parameter is constant regardless of the bulk sample material. [11] When using the ratio of fluorescent X-rays of lead (PbLβ1) and Compton scattering X-rays of rhodium (RhKα) for quantitation, the equation can be given as follows: M f;pbl β1 ¼ a T f;pbl β1 þ c; (4) M s;rhk αc T s;rhk αc where M is the measured intensity, T is theoretical intensity calculated from the function of concentration, and a is a proportionality constant. In addition, c is the intercept, which should theoretically be zero, but c is used for correction of small measurement errors as necessary. f,pblβ1 is the fluorescent X-rays of PbLβ1, and s,rhkαc is the Compton scattering X-rays of RhKα. [7] A typical EDX spectrum is shown in Figure 3. Results and discussions Linearity Table 4 shows the linearity result. The correlation coefficient of all the six elements for the calibration curve meet the acceptance criteria. Overlap correction is used for As because the energies of AsKα (10.53 kev) and PbLα (10.55 kev) are very close. The linearity for the BG-FP method cannot be evaluated directly because the BG-FP method uses only one sample for making each element calibration. Therefore, the standard water solution used in making the element samples for calibration curves was quantitated using the BG-FP method, and correlation was checked between the calibration curve and BG-FP methods for all the six elements. [12] The result for the BG-FP method shows good linearity with the Table 4. Result of linearity Element Cd Pb As V Co Ni R of calibration curve R between calibration curve and BG-FP Note. BG-FP = background fundamental parameters. R means correlation coefficient. calibration curves. However, the calibration curve for Ni has a bigger y-intercept than the other calibration curves due to contamination. In this case, it is not a problem to use the calibration curve method because the calibration curve method can be corrected by the y-intercept; however, the BG-FP method has no y-intercept and cannot be easily corrected. It is necessary to consider eliminating the remaining Ni contamination or including the y-intercept as a constant coefficient in the BG-FP method. The calibration curves for Pb, Cd, Co, and Ni are shown in Figure 4. Accuracy/specificity Table 5 shows the Accuracy/specificity results. In this study, the cellulose samples were quantitated by calibration using water solution sample(s). The accuracy/specificity is based on the recovery rate, which is the percentage of the calculated concentration in comparison to the prepared verification sample concentration. The accuracy/specificity of all the six elements meets this acceptance criteria in both calibration curve and BG-FP methods. However, Ni shows a different tendency compared to the other elements due to contamination from the detector. Quantitation limit The detection limit is calculated as the standard deviation of the intensity for 10 measurements of a blank cellulose sample. The quantitation limits for the elements are shown in Table 6 for measurement times of 300 and 1,200 s with specification limits in the bottom two rows. There is no any specific definition of the term specification in USP <735>; however, in this study, the specification limit is defined as the PDE limit from USP <232>. The two values of specification limits shown are from conversion methods Options 1 and 2a, respectively. From these result, Cd and Pb do not meet the 50% specification at 10 g maximum daily intake even at the measurement time of 1,200 s. The required measurement time is 37,600 s for Cd and 4,800 s for Pb to meet the 50% specification at 10 g maximum daily intake. This measurement time is not practical, especially for Cd; however, this quantitation limit can meet the 50% specification at 1 g maximum daily intake, and the EDX system can perform screening measurements for impurity control in these cases. Robustness USP <735> described the reliability of an analytical measurement should be demonstrated by deliberate changes to experimental parameters. There is no any specific definition 385 X-Ray Spectrom. 2017, 46, Copyright 2017 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/xrs

5 H. FURUKAWA ET AL. Figure 4. Calibration curves for Pb, Cd, Co, and Ni Table 5. Result of accuracy/specificity Sample Item Cd Pb As V Co Ni 5 μg/g cellulose by calibration curve Quantitative value (μg/g) Accuracy (%) μg/g cellulose by BG-FP Quantitative value (μg/g) Accuracy (%) μg/g cellulose by calibration curve Quantitative value (μg/g) Accuracy (%) μg/g cellulose by BG-FP Quantitative value (μg/g) Accuracy (%) Note. BG-FP = background fundamental parameters. Table 6. Result of quantitation limit Quantitation limits (μg/g) Time (s) Method Cd Pb As V Co Ni 300 Calibration curve BG-FP Calibration curve BG-FP % specification at 10 g maximum daily intake 50% specification at 1 g maximum daily intake a Note. BG-FP = background fundamental parameters. Bold face means the quantitation limit does not meet the above both 50% specification. Underline means the quantitation limit does not meet the 50% specification at 10 g maximum daily intake. wileyonlinelibrary.com/journal/xrs Copyright 2017 John Wiley & Sons, Ltd. X-Ray Spectrom. 2017, 46,

6 387 Impurity analysis in drug materials Table 7. Result of robustness Sample Method Quantitation value (μg/g) Cd Pb As V Co Ni 10 μg/g cellulose 2.0 g Calibration curve BG-FP μg/g cellulose 0.1 g Calibration curve BG-FP Fluorescence X-ray energy (kev) Scattering energy for correction (kev) Difference of energy between fluorescence and scattering (%) a Note. BG-FP = background fundamental parameters. of the term experimental parameters in USP <735>; however, in this study, the experimental parameter was defined as sample volume. Table 7 shows the robustness result for two sample volumes. Robustness for As, Co, and Ni does not meet the acceptance criteria. Table 7 shows also the difference of energy between fluorescence and scattering in the bottom row. The energy differences for As, Co, and Ni are bigger than for the other elements. For these cases, correction may not be accurate because the thickness effect is different between the fluorescence and scattering X-rays due to the large energy difference between them. Conclusion EDX is expected to be applied to the screening and control of impurities for the most important risk assessment elements such as Cd, Pb, As, V, Co, and Ni in the pharmaceutical field. EDX reduces the need for complex chemical analysis procedures requiring timeconsuming chemical pretreatment processes. Furthermore, the BG- FP method can reduce the number of standard samples needed for calibration. There is currently a problem about correlation between the calibration curve and BG-FP methods for accuracy/specificity. Results between the calibration curve and BG-FP methods for Cd, Pb, and As are very similar and can be accounted for by normal experimental variation. However, the results for V, Co, and Ni show differences between the calibration curve and BG-FP methods. One reason may be the influence of the different means of correction of matrix effects for the two methods, as for these lower atomic number elements matrix effects are stronger. There is also currently a problem about correction conditions for robustness, and it is necessary to perform further research for more suitable scattering energy conditions. It is also necessary to research the applicability of other matrix materials for this method. References [1] United States Pharmacopeia (2014). General chapters USP <735> X-ray fluorescence spectrometry. [2] United States Pharmacopeia (2013). USP <232> Elemental Impurities Limits. [3] International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human use (2014). Guideline for Elemental Impurities Q3D. [4] Product Information of EDX-7000/8000 on Shimadzu Europe GmbH, 2016; [5] I. Shiraiwa, N. Fujino, J Appl Phys 1966, 5, 886. [6] I. Campbell, Y. Xiao, B. Vrebos, L. Kempenaers, D. Coler, K. Macchiarola, White paper/application notes by PANalytical, 2012; americanpharmaceuticalreview.com/1504-white-papers-application- Notes/ The-Use-of-EDXRF-for-Pharmaceutical-Material- Elemental-Analysis/ [7] H. Ochi, S. Watanabe, Adv X-Ray Chem Anal, Japan 2006, 37, 45. [8] R. Ogawa, H. Ochi, M. Nishino, N. Ichimaru, R. Yamato, X-Ray Spectrom 2010, 39, [9] R. B. Kellogg, Adv X-ray Anal 1984, 27, 441. [10] K. K. Nielson, Adv X-ray Anal 1979, 22, [11] R. Ogawa, H. Ochi, M. Nishino, N. Ichimaru, R. Yamato, Adv X-Ray Chem Anal, Japan 2011, 42, 315. [12] H. Furukawa, N. Ichimaru, S. Watanabe, K. Suzuki, M. Nishino, H. Ochi, D. Davis, Adv X-ray Anal 2016, 59, X-Ray Spectrom. 2017, 46, Copyright 2017 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/xrs