Development Studies of a New Metronidazole Certified Reference Material

Este artigo apresenta os resultados dos estudos realizados com o candidato a material de referência certificado (MRC) de metronidazol, primeiro MRC deste fármaco disponibilizado no mercado e segundo MRC de fármacos brasileiro. Os estudos incluem a determinação dos teores de impurezas orgânicas, inorgânicas e voláteis, validação do método analítico de cromatografia líquida de alta eficiência (CLAE), estudos de estabilidade em condições de transporte e estocagem, estudo de homogeneidade, determinação do teor de metronidazol por balanço de massa, confirmação deste valor por calorimetria exploratória diferencial (DSC) e cálculo de incertezas de medição.


Introduction
Metronidazole (Figure 1) is a 5-nitroimidazole primarily used against protozoal parasites (e.g., Giardia lamblia, Entamoeba histolytica, Trichomonas vaginalis) and anaerobic bacterial infections (e.g., Gardnerella vaginalis). 1 It is also used in the treatment of gastritis caused by Helicobacter pylori 2 and for relief of the inflammatory lesions and erythema of rosacea. 3Its main pharmaceutical finished products for human use are tablets (oral therapies), as well as gels, creams, or lotions (topical or vaginal therapies). 3,4n order to guarantee the quality of pharmaceuticals, the industry and governmental laboratories rely on several tools, including the quality control of raw materials and finished products.For quality control, reference materials (RMs) are highly important, as are the validation of analytical methods and equipment calibration and verification.
Unlike typical RMs, certified reference materials (CRMs) are capable of ensuring metrological traceability of measurement results to the international system of units (SI), which corresponds to an "unbroken chain of calibrations, each contributing to the measurement uncertainty". 5This traceability is a requirement of ISO/IEC 17025:2005; 6 therefore, accredited laboratories rely on the use of CRMs and other tools to ensure it.Through the use of CRMs, measurement results can be meaningfully compared "over time and place". 5he use of CRMs is a common practice in several fields.However, in the pharmaceutical area, the use of reference materials (RM) is much more common; these are typically called reference standards. 7It was only in 2008 that the United States Pharmacopeial Convention (USP) began to produce CRMs of active pharmaceutical ingredients (APIs).Until the present year (2011), only four CRMs, namely dextromethorphan hydrobromide, carbamazepine, phenytoin, and theophylline, became available in their catalogue, 7,8 even though the release of 2500 reference standards had been announced in 2010. 9ecently, the National Metrology Institute of Japan (NMIJ) made available the CRM for 17β-estradiol for clinical chemistry, 10,11 the National Metrology Institute of China (NIM) published the development of a paclitaxel CRM, 12 and another group developed the anthraquinone CRM in Rhei (Chinese herbal medicine). 13n Brazil, national reference standards are produced by the Brazilian Pharmacopeia and the import of USP and European Pharmacopoeia (EP) reference standards is also a common practice.The first Brazilian CRM of an API, the captopril CRM, was made available on the market in 2010 by the National Institute of Metrology, Standardization and Industrial Quality (Inmetro). 14,15he requirements for production of CRMs are given by the ISO Guides 34:2009 16 and 35:2006 17 and include the steps of characterization, short-and long-term stability studies, homogeneity evaluation, and uncertainties estimation.The characterization, in the case of APIs, results in the real analyte concentration in the CRM, which will be reported in a certificate with its measurement uncertainty.The analyte content can be obtained by using methods that directly give this value or else by indirect approaches, like the mass balance calculation, which is based on a difference (100% − Σ impurities). 18Besides giving important information about the analyte impurity profile, 19 the mass balance results in smaller uncertainties of the property value, even if larger uncertainties are obtained for the individual impurities. 18,20This was the approach used here for the certification of the metronidazole CRM and the certified property value was cross-checked by differential scanning calorimetry (DSC).
This paper details all of the studies performed with the Inmetro first batch of the candidate CRM of metronidazole.
The differential scanning calorimetry measurements were performed using a DSC Q2000 (TA Instruments, New Castle, USA) and processed with TA Universal Analysis 2000 software.The DSC cell was calibrated with a sapphire disk (TA Instruments) (baseline calibration) and indium SRM 2272 from NIST (enthalpy and temperature calibration).The purity determination was performed according to method A of ASTM Standard E 928-08 21 and consisted of heating 1-3 mg samples, hermetically sealed in an aluminum pan, at 0.5 ºC min -1 under dynamic nitrogen gas-flow (50 mL min -1 ).
All sample solutions were prepared gravimetrically using an analytical balance (Shimadzu), model AUW 220D, with resolution of 0.01 mg.

Reagents and chemicals
The studies were carried out using the Inmetro first batch of the metronidazole candidate CRM.After satisfactory results in the tests according to the Brazilian Pharmacopeia IV monograph, 22 amber glass flasks were filled with 525 mg metronidazole (nominal mass: 500 mg, excess: 5%) under controlled temperature and humidity conditions, closed with rubber stoppers and aluminum seals and identified with the CRM name and a sequential number.All of the tests were performed with the material packed as sent to the customer.The CRM was dried at 105 °C for 2 h prior to use in the HPLC related substances test, while it was used without prior drying for the homogeneity and stability studies.
A USP reference standard of metronidazole (batch J1l272, 100.0 %) and EP reference standard of metronidazole impurity A (batch 2.1, 100.0%) were also analyzed by HPLC for comparison.The metronidazole USP reference standard was dried at 105 °C for 2 h prior to

Candidate CRM characterization
The content of API in the Inmetro first batch of metronidazole candidate CRM was determined using the mass balance approach according to equation 1 (Table 1).In order to use this equation, it is essential to determine with accuracy the content of organic, inorganic, and volatile impurities.
The organic impurities were determined by HPLC-DAD based on the related substances test in the European Pharmacopoeia V monograph. 23Solutions containing 500, 5, and 0.5 µg g -1 of metronidazole (previously dried) in mobile phase were prepared in triplicate from each candidate CRM flask, and each solution was injected three times into the HPLC system.This test was repeated on three different days.The content of each organic impurity in the CRM samples was determined by peak area ratio according to equation 2 (Table 1).The inorganic impurities were determined by residue on ignition (sulfated ash) and the volatiles by loss on drying (105 °C per 2 h) according to the Brazilian Pharmacopeia IV. 22 The loss on drying results were compared to the water content determined by Karl Fischer coulometric titration (oven method).

HPLC-DAD method validation
The method specificity was evaluated by analysis of the diluent (mobile phase), metronidazole impurity A EP reference standard (5 and 0.5 µg g -1 solutions), metronidazole USP reference standard (500, 5 and 0.5 µg g -1 ), metronidazole CRM samples (500, 5 and 0.5 µg g -1 ), and mixtures of metronidazole CRM and metronidazole impurity A EP reference standard (500 and 5 µg g -1 , respectively).Additionally, samples of metronidazole CRM were exposed to the following stress conditions: 170 °C for 1 h, UV at 254 nm for 24 h; 1 mol L -1 HCl, 1 mol L -1 NaOH, or 3% H 2 O 2 for 1 h.After treatment, the samples were neutralized (in case of acidic or basic conditions) and diluted with mobile phase to 500 and 5 µg g -1 .
In order to evaluate the metronidazole stability in the diluent (mobile phase), 500 and 5 µg g -1 solutions of metronidazole CRM were transferred to HPLC vials, kept at 15 °C in the auto-sampler, and injected three times into the HPLC at 0, 5.7, 11.8, 17.8, 32.7, 53.6, and 80.5 h.
For linearity evaluation, 16 solutions of metronidazole USP reference standard or metronidazole impurity A EP reference standard in the concentration range between 500 and 0.01 µg g -1 were analyzed.The limit of quantitation (LOQ) and the limit of detection (LOD) were estimated from the linearity results in the range between 1.0 and 0.010 µg g -1 , which was appropriate for observation of responses (peak heights) equivalent to 10 or 3 times the noise (LOQ and LOD, respectively), as required by the signal-to-noise (S/N) ratio method. 24Only results above LOQ were considered in the regression analysis for linearity evaluation. ( *[org] = content of each organic impurity (%); A imp = peak area of each organic impurity (500 µg g -1 solution); A metr dil = metronidazole peak area (5 µg g -1 solution); DF = dilution factor (500 to 5 µg g -1 ); ΣA imp = sum of peak areas of organic impurities; u org = combined standard uncertainty of organic impurities content (%); ∂ org /∂ xi = partial differential of organic impurities content with respect to each uncertainty component x i ; u xi = standard uncertainties of input estimates x i (%); u c = combined standard uncertainty (inorganics or volatiles) (%); SD = standard deviation (%); n = number of measurements; x = mean of results (%); U = expanded uncertainty of analytical balance (mg); k = coverage factor; m = average weighed mass (mg); u bb = uncertainty due to between bottle (in)homogeneity (%); MS between = mean square between groups; MS within = mean square within groups; n = number of replicates; u stab = uncertainty due to stability (u sts : short-term, u lts : long-term) (%); s (b1) = slope uncertainty (% day -1 ; % week -1 ); t = time (days; weeks); u CRM = combined standard uncertainty of property value (%); T S = sample melting temperature (K); T 0 = pure material melting temperature (K); R = gas constant (J mol -1 K -1 ); c = impurity mole fraction; DH = molar enthalpy of fusion (J mol -1 ); F = molten material fraction.
The repeatability was evaluated by six repeated injections of metronidazole CRM (500, 5 and 0.5 µg g -1 ).This test was repeated on a different day by a different analyst to verify the intermediate precision.The recovery tests for accuracy (and also specificity) evaluation were performed using solutions of metronidazole impurity A EP reference standard (2.5, 5.0 and 10.0 µg g -1 ), with or without addition of 500 µg g -1 of metronidazole CRM.This CRM was used as a matrix and was individually analyzed to determine the metronidazole impurity A content, which was subtracted before recovery calculation.

Homogeneity testing, short-and long-term stability studies
The between-bottle variation was evaluated using 13 candidate CRM flasks selected at random and analyzed in triplicate.The HPLC analysis was done as previously described, the content of organic impurities was calculated by equation 2 (Table 1) and the metronidazole content was determined by mass balance according to equation 1 (Table 1).
For the short-term stability study, 10 CRM flasks selected at random were exposed simultaneously to the temperature of 50 °C.On pre-determined days (1, 4, 8, 10, 15, 29, 31, 52, 64, and 66), one flask was taken out of the oven and individually analyzed (classical design).Solutions containing 500, 5, and 0.5 µg g -1 were prepared in triplicate from each CRM flask, and each solution was injected three times into the HPLC system.Three other CRM flasks, kept at the reference temperature of 25 °C and analyzed in the same way, were used as controls (t = 0).The metronidazole content was calculated by mass balance according to equation 1 (Table 1).
For the long-term stability study, CRM samples were kept at the recommended storage conditions (25 ºC) and analyzed four times throughout the year of certification studies, namely in the 1 st , 25 th , 34 th , and 42 th weeks.At each time, three flasks were analyzed in triplicate.After conclusion of this study, a monitoring program of the CRM samples was started.

Estimation of measurement uncertainties
The uncertainty estimation was performed according to the Eurachem/Citac Guide 26 and the relevant uncertainty sources can be seen in the cause and effect diagram (Figure 2).Based on equation 1 (Table 1), the uncertainties due to the determination of organic, inorganic, and volatile impurities were included.The combined standard uncertainty of the organic impurities content (u org ) was calculated by equation 3 (Table 1).The considered uncertainty sources were essentially the variables of equation 2 (Table 1), but possible differences in specific absorptivities of metronidazole and its impurities were also taken into account.The combined standard uncertainties (u c ) of the contents of inorganic or volatile impurities (u inorg and u vol , respectively) were individually calculated by equation 4 (Table 1).Additionally, the uncertainty due to between-bottle (in) homogeneity (u bb ) was calculated by equation 5 (Table 1), while the short-and long-term stability uncertainties (u sts and u lts , respectively) were estimated using equation 6 (Table 1), according to the ISO Guide 35:2006. 17As will be explained later, u sts was not taken into account in the calculation of the combined standard uncertainty of the CRM property value (u CRM ), which was carried out using equation 7 (Table 1) according to the law of propagation of uncertainties 26 and verified by the Kragten method, which is also derived from this law, but is based on the separation of the contribution of different error sources. 26,27Finally, the CRM expanded uncertainty (U) 26 was calculated by equation 8 (Table 1).

Results and Discussion
The certification steps were: (i) HPLC-DAD method validation; (ii) CRM characterization (determination of contents of organic, inorganic, and volatile impurities); (iii) homogeneity testing; (iv) short-term stability studies; (v) long-term stability studies; and (vi) property value assignment by mass balance.Whenever applicable, the uncertainties were estimated.

HPLC-DAD method validation
The determination of related substances (organic impurities from synthesis or degradation products) of metronidazole was performed by HPLC-DAD according to the European Pharmacopoeia V monograph. 23The Brazilian Pharmacopeia IV 22 and the United States Pharmacopeia 32 28 recommend the use of thin layer chromatography for evaluation of metronidazole related substances, which was not appropriate for certification purposes.
The validation results of the HPLC-DAD method are summarized in Table 2, while Figures 3-5 show chromatograms of metronidazole CRM, metronidazole USP reference standard, and metronidazole impurity A EP reference standard, respectively.In the specificity tests, from all tested stress conditions (170 °C for 1 h; UV 254 nm for 24 h; 1 mol L -1 NaOH, 1 mol L -1 HCl, or 3% H 2 O 2 for 1 h), only the basic pH condition resulted in significant changes in the metronidazole content (reduced from 99.60% to 96.03%) and in the impurity profile (new impurity peaks, increase in peak areas of some impurities).However, the metronidazole peak was still well resolved from the impurity peaks and its peak purity index was preserved (0.9996), demonstrating the method specificity.Even though the metronidazole instability upon exposure to light has been reported, 22 this effect was not observed.
The linearity was evaluated in the range between LOQ and 500 µg g -1 , and also between LOQ and 10 µg g -1 , which is the most probable concentration range for impurities, considering the limits of 0.5 µg g -1 for each impurity and 1.0 µg g -1 for total impurities (0.1% and 0.2%, respectively, of 500 µg g -1 ). 23The regression analysis resulted in squared correlation coefficients (r 2 ) of at least 0.9996 for metronidazole and 1.0000 for metronidazole impurity A. Therefore, the method linearity 29 was demonstrated.
The repeatability results indicated a smaller standard deviation of normalized peak areas for the 5 µg g -1 solutions compared to the 0.5 µg g -1 solutions.Therefore, the variables of equation 2 (Table 1) related to the diluted solution were associated with the 5 µg g -1 concentration.
For robustness evaluation, eight HPLC analyses were performed with variations in mobile phase flow-rate (variables A/a), detection wavelength (variables B/b), KH 2 PO 4 concentration (variables C/c), and methanol concentration: (variables D/d).From the results of the eight experiments (s: 99.5986, t: 99.6026, u: 99.5878, v: 99.5996, w: 99.5925, x: 99.5954, y: 99.5945, and z: 99.5974 %), the values of the variables were determined to be A = 99.5972, a = 99.5950,B = 99.5973,b = 99.5948,C = 99.5934,c = 99.5978,D = 99.5983, and d = 99.5938%.According to Bedegral et al., 25 if the differences (absolute values) between the results for the upper case variable (nominal condition) and their corresponding lower case variable (departure condition), for instance |A-a|, are not larger than s√2, where s is the standard deviation of results, the variable has an insignificant effect and the method is not sensitive to changes in the variable concerned.The same is valid for |B-b|, |C-c| and |D-d|.In our study, all of the differences (Table 2) were smaller than s√2 = 0.0065% (calculated from s = 0.0046%) and the method was demonstrated to be robust for all evaluated parameters.
The results for the other validation parameters (stability in solution, LOQ, intermediate precision, and accuracy) (Table 2) were also satisfactory.Therefore, the HPLC method was considered validated.

Determination of organic impurities
The content of organic impurities was calculated by peak area ratio according to equation 2 (Table 1) considering that (i) it is not an assay calculation and does not rely on the use of CRMs of metronidazole and metronidazole impurity A, which are not available and therefore could not be used to guarantee metrological traceability; and (ii) the peak area of metronidazole and impurities are in the same order of magnitude, since the peak area of metronidazole   .Experimental conditions: same as in Figure 3. is obtained from the 5 µg g -1 solution and corrected by the dilution factor (DF).
Table 3 shows the related substances contents from three HPLC determinations, with calculations based on equation 2 (Table 1) and also on the assay against USP metronidazole reference standard (as standard for unknown impurities) and EP metronidazole impurity A reference standard.An agreement between the peak area ratio method and assay results was observed.
The average organic impurity content by peak area ratio was 0.0676%, from which 0.02% corresponds to metronidazole impurity A and the other part to unknown impurities.These results are in accordance with the acceptance criteria of the European Pharmacopoeia V monograph 23 (maximum of 0.1% of each organic impurity and 0.2% of total organic impurities).The largest combined uncertainty (0.0044%, day 1) was considered for u CRM calculation.

Determination of inorganic and volatile impurities
The content of inorganic impurities determined by residue on ignition was 0.0455% (n = 19), which is in accordance with the specified limit of 0.1% in the Brazilian Pharmacopeia IV monograph for metronidazole. 22The calculated value of u inorg according to equation 4 (Table 1) was 0.0078% (considering m = 503.76mg, U = 0.04 mg and k = 2).
The volatiles content determined by loss on drying was 0.260% (n = 16), which is also in accordance with the specified limit of 0.5% in the Brazilian Pharmacopeia IV monograph. 22The combined standard uncertainty (u vol ) estimated by equation 4 (Table 1) was 0.013% (m = 1010.45mg, U = 0.04 mg and k = 2).The water content determined by the Karl Fischer coulometric titration was 0.0245% (n = 10, u = 0.0004%), which was a smaller value compared to the result from the loss on drying test.This difference may be explained by the fact that the Karl Fischer method is insensitive to residual solvents; therefore, the Karl Fischer and loss on drying results may not be comparable. 30mogeneity testing, short-and long-term stability studies In the homogeneity testing, the single-factor analysis of variance of results for the 13 flasks analyzed in triplicate resulted in a mean square between groups (MS between ) of 2.7624 × 10 -5 and a mean square within groups (MS within ) of 2.1054 × 10 -6 .Using equation 5 (Table 1), the uncertainty due to between-bottle (in)homogeneity (u bb ) was estimated as 0.0029%.The short-term stability studies were carried out to evaluate the CRM stability under transport conditions. 17he CRM flasks were exposed at the same time to the temperature of 50 ºC and then analyzed individually on predetermined days.This design, which is called "classical", is performed under reproducibility conditions and therefore is expected to result in larger uncertainties compared to the "isochronous" design, performed under repeatability conditions (each flask is introduced into the oven on a pre-determined day and then all flasks are analyzed at the same time at the end of the study).However, the classical design was still considered more appropriate here, since the results are made available throughout the study.
The metronidazole contents during the 66 days of the short-term stability study are shown in Figure 6.The regression analysis of the curve resulted in an F value smaller than F crit (3.23 < 3.92) and a p-value larger than 0.05 (0.07); therefore, the curve slope was considered insignificant and the CRM stability under transport conditions (50 ºC, 66 days) was confirmed.Using equation 6 (Table 1), based on the value of s (b1) = 1.48 × 10 -5 (regression analysis) and on the 66-day period, the short-term stability uncertainty (u sts ) was estimated as 0.0010%.Since this value was smaller than the long-term stability uncertainty (see next paragraph), it was not considered relevant as an uncertainty component 17 and was not included in the u CRM calculation.
The long-term stability studies were carried out to evaluate the stability of the metronidazole candidate CRM under storage conditions (25 °C). Figure 7 shows the metronidazole contents monitored throughout 42 weeks.

CRM certified property value and uncertainty
The certified value of the metronidazole CRM was determined by mass balance according to equation 1 (Table 1).Since metronidazole was dried prior to characterization and has to be dried by the consumer prior to use, as stated in the CRM certificate, the volatiles content was considered to be zero in the mass balance calculation and the uncertainty of volatiles content was disregarded in the CRM combined standard uncertainty calculation (u CRM ).It should also be noted that the peak area ratio method for determination of organic impurities leads to small differences compared to the methods for quantification of inorganic and volatile impurities, since it deals only with organic compounds (analyte and impurities).However,  since the volatiles were removed from the samples prior to characterization and the inorganic impurities content was very small, the mentioned differences were not considered to be significant.
As shown in Table 4, the certified property value calculated by mass balance was 99.887%, the CRM combined standard uncertainty (u CRM ) estimated by equation 7 (Table 1) was 0.007% and the CRM expanded uncertainty (U CRM ) calculated according to equation 8 (Table 1), using a value of 2 as the coverage factor k for a confidence level of approximately 95%, 31 was 0.015%.The uncertainty u CRM due to uncertainties of input parameters x 1 … x n was validated by a spreadsheet method described by Kragten, 27 based on the separation of uncertainty sources.
The certified value was cross-checked by differential scanning calorimetry (DSC), according to the recommendation for use of one primary method or two independent secondary methods for reference material certification. 17DSC has been claimed as a "primary method", which refers to a method that does not rely on the use of "measurement standard for a quantity of the same kind". 32In practice, however, empirical corrections are needed, which "reduce the method's claim to produce traceable results". 33DSC is a "classical thermal analysis technique" 18 based on the Van't Hoff law and its derived equation (equation 9, Table 1), which infers that the presence of impurities leads to a depreciation of the melting point. 21The DSC results for metronidazole (99.82 ± 0.01% mol/mol, previously dried sample) validated the mass balance result.The equivalence between the purity determined by DSC and the content determined by mass balance/HPLC had already been reported for highly pure substances. 34,35he CRM metronidazole content was also determined by titration in non-aqueous media with perchloric acid, according to the assay in the Brazilian Pharmacopeia IV monograph, 22 and the results were in the range of 99.28% to 100.52%.This large variability of titrimetric method results is not unusual. 36In addition, metronidazole is a 5-nitroimidazole compound, i.e., it contains a NO 2 group on the 5 th position of the imidazole ring 1,37 (Figure 1).All of the metronidazole impurities described in the literature 38 also have the imidazole group, which is reported to react in a 1:1 molar ratio with perchloric acid; 39 therefore, they can react with the titrant.Due to the non-specificity of this technique 33,40 and its large variability, the titration was less appropriate for certification purposes compared to the mass balance approach, even if its use would be recommended as a primary direct method. 17,33,40

Conclusions
The Inmetro first batch of the candidate certified reference material of metronidazole, the first CRM of this active pharmaceutical ingredient, has been certified.The certified property value and the expanded uncertainty (U), using the value of 2 as the coverage factor for a confidence level of approximately 95%, was (998.887 ± 0.015) mg g -1 (previously dried sample).
The metronidazole CRM is an important tool for ensuring metrological traceability of measurement results to the SI by analytical laboratories and research institutes, in compliance with ISO/IEC 17025:2005.

Figure 2 .
Figure 2. Cause and effect diagram with uncertainty sources for metronidazole candidate CRM.

Figure 4 .
Figure 4. HPLC-DAD chromatogram for the related substances test of the metronidazole USP reference standard (500 µg g -1 ) (a) and enlargement (b), showing the peaks of metronidazole (t R = 6.8 min), metronidazole impurity A (t R = 5.4 min) and unknown impurities (t R = 4.6 min and 10.2 min).Experimental conditions: same as in Figure 3.

Figure 5 .
Figure5.HPLC-DAD chromatogram for the related substances test of the metronidazole impurity A EP reference standard (5 µg g -1 ).Experimental conditions: same as in Figure3.

Figure 6 .
Figure 6.Results for the short-term stability study of metronidazole candidate CRM.

Figure 7 .
Figure 7. Results for the long-term stability study of metronidazole candidate CRM.

Table 1 .
Equations used in the certification studies of the metronidazole candidate CRM

Table 2 .
Validation results of HPLC-DAD method for determination of metronidazole related substances

Table 3 .
Related substance content of metronidazole candidate CRM, determined by peak area ratio and by assay against USP and EP standards

Table 4 .
Metronidazole content of the certified reference material and uncertainties a Volatiles were disregarded, since the CRM has to be dried prior to use.b Combined standard uncertainty.c Expanded uncertainty (95% confidence level, k = 2).Vol. 23, No. 3, 2012