2736 Eliangiringa Kaale1,2 1 ¨ Ludwig Hollein Ulrike Holzgrabe1 1 University
¨ of Wurzburg, Institute of Pharmacy and Food ¨ Chemistry, Wurzburg, Germany 2 School of Pharmacy, Muhimbili University of Health and Allied Sciences, Dar Es Salaam, Tanzania
Received January 19, 2015 Revised April 21, 2015 Accepted April 21, 2015
Electrophoresis 2015, 36, 2736–2744
Research Article
Development and validation of a generic stability-indicating MEEKC method for five fluoroquinolone antibiotics In this paper, a reliable stability-indicating generic MEEKC method for the analysis of five commonly used fluoroquinolones (FQs) has been developed and optimized by a central composite circumscribed experimental design. The separation was carried out using a fused silica capillary (60.2 cm total length) and a microemulsion (ME) composed of 81.75% (w/w) of a 125 mM NaH2 PO4 solution having a pH of 2.75, 2.65% (w/w) SDS, 1.00% (w/w) n-octanol, 6.60% (w/w) n-butanol and 8.00% (w/w) 2-propanol. A voltage of 28 kV was applied in a reverse polarity mode. A linear relationship was established from 0.04 to 0.48 mg/ml with R2 values higher than 0.98 for all five FQs. Both repeatability and intermediate precision were less than 3% and accuracy ranging from 97 to 100%. Of note, ciprofloxacin impurity A and ofloxacin impurity A could be separated from the respective drug substance in a single run which cannot be achieved using the official HPLC method from the European Pharmacopoeia. Forced degradation of all FQs under heat, in acidic and alkaline medium, in the presence of oxidizing agents and under neutral hydrolysis conditions was investigated. The highest yield of degradation products was observed using oxidative hydrogen peroxide. Hence, the proposed MEEKC method can be used for the quantitative determination of the five FQs and their potential impurities within a total runtime of 20 min. Keywords: Fluoroquinolone antibiotics / Impurity profiling / MEEKC / Stability-indicating method DOI 10.1002/elps.201500025
Additional supporting information may be found in the online version of this article at the publisher’s web-site
1 Introduction Fluoroquinolones (FQs) belong to an important group of antibiotics which are commonly used in human and veterinary medicine for the treatment of a wide variety of infections. Their activity is based on the interference with various topoisomerases [1]. This makes them broad-range antibiotics
Correspondence: Professor Eliangiringa Kaale, School of Pharmacy, Muhimbili University of Health and Allied Sciences, Dar Es Salaam, Tanzania, P.O. Box 65545/11103, Upanga West, Dar Es Salaam, Tanzania. E-mail: [email protected]
Abbreviations: API, active pharmaceutical ingredient; CCD, central composite design; CFX, ciprofloxacin; FQ, fluoroquinolone; IPA, isopropanol; LFX, levofloxacin; LoFX, lomefloxacin; ME, microemulsion; MFX, moxifloxacin; NFX, norfloxacin; OFX, ofloxacin; PDA, Photodiode Array Detector; Ph. Eur., European Pharmacopoeia; SC, sodium cholate; TEA, trietylamine
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
which are effective against gram-positive and -negative bacteria [2–4]. Usually the quality of commercially available finished pharmaceutical products is determined by several critical quality attributes [5]: (i) identity, (ii) assay, (iii) disintegration, (iv) dissolution and (v) impurity profile [6,7]. Developing separation methods which are able to determine the related substances of a compound is a major issue during and after drug development. Routinely applying such methods still remains a challenge for quality control laboratories particularly in developing countries with restricted resources. In many cases, compendial methods are cumbersome or tedious and often require the application of more than one technology in order to be able to resolve all related substances [8–10]. However, the availability of suitable analytical methods which are able to separate an active pharmaceutical ingredient (API) from all its potential impurities is paramount during the quality assessment of pharmaceutical products and the detection of counterfeits [11]. For the recent ten years several analytical protocols for the determination of commonly available FQs have
www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2015, 36, 2736–2744
2737
Figure 1. Chemical structures of the investigated FQs and the internal standard lomefloxacin.
been published such as UV spectroscopy for quantitative analysis [12, 13], or TLC and high-performance thin layer chromatography (HPTLC) methods [14–17] as well as stability-indicating HPLC methods [18–22] and CZE, respectively [23]. The monographs which have been added to the European Pharmacopoeia (Ph. Eur.) describe LC tests for the control of related substances of the most important FQs. Nevertheless, it is not always possible to resolve all potential impurities using one technology only, for example in the case of ciprofloxacin (CFX), norfloxacin (NFX) and ofloxacin (OFX). The respective monographs which can be found in the Ph. Eur. and the United States Pharmacopoeia (USP) describe the combination of TLC and LC in order to determine the related substances of the compounds [8–10, 24–26]. Enantioselective methods for OFX [27, 28] and moxifloxacin (MFX) [29] have also been published; all of them apply CZE in combination with a chiral selector in order to resolve the respective enantiomers. On the other hand, simpler methods for determining known and/or unknown impurities are available for the analysis of CFX [30, 31]. Microemulsions (MEs) have been described for the first time in 1959 [32] and prove to be a potential methodology for separating an array of compounds with diverse physicochemical properties [33, 34]. They are isotropically clear solutions consisting of a lipophilic component (e.g. octane) and water, and are stabilized by adding a surfactant or a co-surfactant whose presence reduces the interfacial tension to almost zero [35, 36]. Today ME systems are widely applied in CE in a mode called MEEKC [37] which utilizes MEs as the separation medium [38]. Recently, a generic MEEKC method was successfully developed and applied for the analysis of 15 different antimalarials [39]. This is a great benefit for the practical work particularly in resource-constrained settings because a testing laboratory would only need one inventory of supplies catering for all 15 drug products instead of having 15 inventories for each individual API.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
To the best of our knowledge there is no existing generic MEEKC method for the separation of the commonly used FQs CFX, NFX, MFX, levofloxacin (LFX) and OFX including their potential related substances (Fig. 1 and Supporting Information Fig. S1). Our study aimed at developing and validating a generic MEEKC method for the separation and quantitative determination of these five mostly used FQs by applying an experimental design space approach to optimize the critical separation parameters. In addition, the purity of the single compounds can be analysed as the method is also able to resolve the respective related substances from the main signal when individual runs are performed for each API. During method development commercially available and known impurities which are described in the respective monographs of the Ph. Eur. were considered. An experimental design space approach was applied to optimize the critical MEEKC parameters and to (i) enable the quantitative determination of the compounds, (ii) to detect counterfeits with regard to the exchange of the API and (iii) to separate the respective impurities from the individual APIs and thus determining the purity of the compound.
2 Materials and methods 2.1 Apparatus and electrophoretic conditions A Beckman P/ACE MDQ capillary electrophoresis instrument was applied (Beckman Coulter GmbH, Krefeld, Germany), equipped with a capillary cartridge cooling system and a photodiode array detector (PDA). Data were acquired and processed using the Beckman 32 Karat software (version 4.01). The capillary was 60.2 cm long (50 cm effective length, internal diameter 50 m) and supplied from BGB Analytik (Schloßb¨ockelheim, Germany). The applied voltage was 28 kV (0.17 min ramp time) using reverse polarity. The capillary was thermostated at 25°C and all samples were injected applying a pressure of 0.5 psi for 10 s. The PDA
www.electrophoresis-journal.com
2738
E. Kaale et al.
was set at 295 nm using a data sampling rate of 4 Hz, a bandwidth of 10 nm, and normal filtering, respectively. New capillaries were preconditioned at 40°C by rinsing them with 0.1 M NaOH (10 min) – purified water (5 min) – 0.1 M H3 PO4 (10 min), and purified water (5 min) at a pressure of 40 psi. Between two runs the capillary was flushed using the following reagents at 40 psi and 25°C: 0.1 M NaOH for 5 min, methanol for 3 min, 0.1 M H3 PO4 for 3 min, purified water for 3 min and running buffer for 5 min. The pH values were determined using a calibrated glass electrode from Radiometer Analytical (Villeurbanne, France). All solutions were passed through 0.22 m syringe filters prior to use. The optimization by experimental design and multivariate analysis was carried out using the Modde 4.0 software (Umetrics, Umea, ˚ Sweden).
2.2 Chemicals and materials SDS, Tris-phosphate, sodium cholate (SC), H3 PO4 (85% v/v), NaOH, anhydrous NaH2 PO4 , FeCl3 , triethylamine (TEA), H2 O2 , n-butanol, n-octanol as well as CFX, lomefloxacin (LoFX) and LFX primary standards were obtained from Sigma-Aldrich Chemie (Taufkirchen, Germany). Isopropanol (IPA) and methanol were purchased from Fisher Scientific (Loughborough, UK). OFX and NFX primary standards were supplied through Alfa Aesar GmbH & Co KG (Karlsruhe, Germany). MFX hydrochloride, CFX impurity A, NFX impurities A and E, OFX impurities A and E as well as “Ciprofloxacin hydrochloride for peak identification CRS” and “Norfloxacin for peak identification CRS” were obtained from the EDQM (Strasbourg, France). Two CFX commercial samples manufactured by Riva Pharma, Cairo, Egypt, were bought in Dar Es Salaam, Tanzania. High purity water was prepared by a Milli-Q water purification system (Merck Millipore, Billerica, USA).
2.3 Preparation of the microemulsion The background electrolyte (125 mM NaH2 PO4 ) was prepared as follows: 7.5 g of anhydrous NaH2 PO4 were dissolved in about 400 ml Millipore water, the pH of the solution was adjusted to 2.75 with 1.0 M H3 PO4 and the volume was made up to 500 ml with water. The standard o/w ME was prepared by mixing 0.5 g n-octanol, 3.3 g n-butanol, 1.33 g SDS, 4.0 g of IPA and 40.88 g of the background electrolyte solution. The mixture was either stirred for 1 h or sonicated for 30 min to obtain an optically transparent ME.
Electrophoresis 2015, 36, 2736–2744
MeOH (85:15 v/v). The solution was sonicated for 10 min to facilitate dissolution. Diluent B: 25 mM H3 PO4 and MeOH were mixed in the ratio 85:15 (v/v). For preparing the standard solutions, 50.0 mg of the respective FQ were individually dissolved in 50.0 ml of diluent A, whereas for the market sample stock solutions portions equivalent to 50.0 mg of the respective FQ were weighed from crushed tablets and dissolved in 50.0 ml of diluent A. Standard and sample working solutions were prepared by diluting 10.0 ml aliquots of the respective stock solution to a final volume of 25.0 ml using the same diluent (0.4 mg/ml). For peak identification purposes, portions equivalent to 50.0 mg of the respective FQ were weighed from crushed tablets and dissolved in 50.0 ml of diluent B, and in the case of CFX and NFX, 10.0 mg of “Ciprofloxacin hydrochloride for peak identification CRS” and 10.0 mg of “Norfloxacin for peak identification CRS”, respectively, were individually dissolved in 20.0 ml of diluent B. Electropherograms obtained from these solutions were used to determine the purity of the commercial CFX and NFX tablets by comparison of the individual peak patterns. For the forced degradation studies the test substances were individually dissolved in 0.1 M NaOH, 0.1 M H2 SO4 and a 3% solution of H2 O2 with and without the addition of 1% FeCl3 , respectively, to obtain solutions of 0.4 mg/ml. The alkaline and acidic solutions were kept for 24 h and then refluxed at 100°C for 60 min whereas the peroxide solutions were refluxed at 100°C for 30 min. A thermal degradation test was carried out by dissolving the five APIs in a mixture of 25 mM H3 PO4 and MeOH (85:15 v/v) and refluxing at 100°C for 24 h.
2.5 Method development and optimization by design of experiments (DoE) 50.0 mg of each FQ were dissolved in 50.0 ml of diluent B, and 10.0 ml of this solution were diluted to a final volume of 25.0 ml using the same solvent (0.4 mg/ml). Compound separation and optimization of the method was carried out using this model solution. The separation of the individual FQs from their impurities was investigated by preparing solutions containing the API and their respective impurities (i.e. CFX and impurities A, B, C, D and E; NFX and impurities A, E, H and K; MFX, and OFX and impurities A and E; see Fig 1). The most important factors affecting the separation of the compounds were selected for method optimization by applying a central composite response surface modelling experiment, and from screening trials a total set of 27 experiments was derived by application of a central composite design (CCD, Supporting Information Table S1).
2.4 Sample solution preparation All solutions, unless stated otherwise, were prepared using two different diluents prepared as follows: Diluent A: 50.0 mg of LoFX (internal standard) were dissolved in 500.0 ml of a mixture of 25 mM phosphoric acid and C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
2.6 Method validation Validation was carried out with reference to the methodologies described in the guideline Q2(R1) “Validation of www.electrophoresis-journal.com
Electrophoresis 2015, 36, 2736–2744
Analytical Procedures” of the International Conference on Harmonisation (ICH)., and in the USP [40, 41].
2.6.1 Specificity Spiking FQ samples with available impurity standards (cf. Section 2.5) were applied for compound identification and assessment of selectivity. If impurities described in the Ph. Eur. were not commercially available, electropherograms of the samples were compared to those obtained with standard solutions and the forced degradation studies. Since the excipients used for the formulation of the proprietary products were not known the comparison with a blank run (i.e. a placebo mixture) was not possible.
2.6.2 Linearity and accuracy A stock solution was prepared by weighing amounts equivalent to 50.0 mg of each FQ and individually dissolving them in 50.0 ml of diluent A. Serial dilutions using the same solvent were made to obtain five levels (10, 40, 80, 100 and 120%) by standard addition method. Independent controls were prepared at levels of 80, 100 and 120%. The calibrator (i.e. S1-S5) and control (i.e. C1-C3) solutions were randomized on a sequence as S1-C1-C2-S2-C3-S3-C1-C2-C3-S4-C1-C2-C3-S5 in order to minimize errors due to possible variations of the sample or the ME over time. Calibration curves were constructed by plotting the peak responses vs. the respective concentration. Accuracy was determined by evaluating the mean recovery of the analyte at the concentration levels of 80, 100 and 120% from a spiked sample, i.e. aqueous solutions of the compounds which were extracted from tablets. An ANOVA test was performed in order to investigate the fitness of the calibration model.
2.6.3 Precision Repeatability and intermediate precision were determined by independently preparing six replicate working solutions at concentrations corresponding to a level of 100% as described in Section 2.4. Intermediate precision was assessed using a different technical equipment and a different laboratory day.
3 Results and discussion 3.1 Method development: preliminary trials and optimization of the MEEKC separation conditions by DoE A total of approach n-octanol, respective
four variables was investigated in a multivariate by varying the percentage concentrations of n-butanol, IPA and SDS within the ME (see section). Four the most influential variables were
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
CE and CEC
2739
selected and investigated using CCD at five levels in the following concentration ranges: 1.35–3.95% (w/w) SDS, 4.6–8.6% (w/w) n-butanol, 0.6–1.2% (w/w) n-octanol and 4–12% (w/w) n-octanol, respectively. Table S1 (Supporting Information) gives a summary matrix for the CCD design used in the multivariate analysis. The resolution Rs corresponding to the critical peak pair CFX-NFX, the migration time, the number of theoretical plates and the asymmetry factor of the last migrating peak OFX were selected as response variables determining the separation quality. It was assumed that if the generic method would be able to resolve the five FQs, it could also resolve the related substances from each individual API. This was confirmed through spiking throughout the method development. Optimal separation conditions were derived from preliminary trials where Tris-phosphate, NaH2 PO4 and TEA were used at different concentrations (20–200 mM) and pH values (1.85– 3.50). The results indicated that with TEA and Tris-phosphate buffer no separation between CFX, NFX and OFX could be achieved. However, only MFX was resolved from all other signals. Using a 125 mM NaH2 PO4 buffer at pH 2.75, at least two peaks were baseline separated while CFX and NFX comigrated. As the apparent pH of the ME system was 2.95 and the pKa values of the basic nitrogen atoms ranged between 6 and 9, the amino groups were fully protonated [42]. In reversed polarity mode, one would not expect them to move to the anode considering that the EOF is almost suppressed at the low pH value. With SDS being present in the ME the cationic FQs are incorporated into micelles through ionic interaction, which probably get attracted to the anode due to the overall negatively charged complex. In addition, different types of surfactants (SC, Brij-35, SC+Brij-35, SC+Triton X-100, SDS+SC) were investigated. Of note, neither using an individual surfactant alone nor combining them could deliver better results compared to the application of SDS in the concentration range between 2 and 3.5%. From this experiment, SDS was chosen in all further experiments. In order to investigate the influences of temperature on the separation quality the capillary temperature was varied from 20 to 30°C at an interval of 5°C. At lower temperature the OFX migration time was longer than 40 min, however its peak symmetry and the separation from the other compounds were improved at 25°C. At 30°C the current became too high. Subsequently, the capillary temperature was fixed at 25°C and the influence of various oils (i.e. n-octane, n-octanol, n-heptane and ethyl acetate) was evaluated. Using n-octanol at concentrations between 0.5 and 1.0% yielded an improved separation between CFX and NFX, however the number of theoretical plates decreased in this range. The influence of the commonly used co-surfactant, n-butanol, was also investigated between 4 and 8% (w/w), and both resolution and number of the theoretical plates could be raised to 600 000 for the last migrating peak OFX. Adding small organic modifiers to facilitate solubility and partitioning was also investigated, and 4 to 15% (w/w) of ACN, MeOH as well as IPA, respectively, were added. Using ACN resulted into frequent capillary breaks as it was dissolving the www.electrophoresis-journal.com
2740
E. Kaale et al.
Electrophoresis 2015, 36, 2736–2744
Figure 2. Response surface plots displaying the multifactorial influence of variables on the separation parameters.
polyimide coating within the first two or three runs which is consistent with observations previously published by Baeuml and Welsch [43]. The results indicated that the resolution between the critical peak pair CFX-NFX reached a maximum using 8–10% (w/w) of IPA, whereas at higher amounts (15% (w/w)) it decreased and the analysis time went up to 50 min. Using MeOH instead of ACN and IPA could not improve the separation of the compounds at all. Thus, IPA was selected. The molarity of the phosphate buffer, its pH value and the capillary temperature were fixed at 125 mM, 2.75 and 25°C, respectively. 3.1.1 Influence of n-octanol From the initial trials it was observed that CFX and NFX were migrating closely. This might be due to their similar chemical structures, as the only difference is a cyclopropyl instead of the ethyl substituent on the piperazine ring in position 7 (cf. Fig. 1). However, with an increasing octanol concentration, the resolution between CFX and NFX gradually improved up to 3.2 at 1% (w/w) n-octanol; with higher octanol concentrations no better resolution could be achieved, as it is illustrated in Fig. 2A. The influences were studied at five levels, i.e. 0.6, 0.8, 1.0, 1.2 and 1.4% (w/w). C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The combined multivariate interaction effect with IPA has also shown to have a significantly positive effect in resolving this peak pair (results not shown). n-Octanol had a relatively weak effect on both migration times and asymmetry factors of OFX.
3.1.2 Influence of n-butanol The influence of n-butanol in the ME was studied at 4.6, 5.6, 6.6, 7.6 and 8.6% (w/w), respectively, and a positive effect of using it in the ME could be observed even at low concentrations. A steadily increase of the number of theoretical plates was found up to a percentage of 6.6% (w/w). A further increase resulted into a gradually decrease of the peak efficiency, in a prolonged migration time and a large asymmetry of the OFX peak. The multivariable response surface plot is depicted in Fig. 2D. A significant negative multivariate interaction effect was found with IPA, thus combining a high amount of n-butanol and IPA as well as lowering the concentration of SDS resulted into a longer analysis time. Two experiments with this combination were removed from the model because of the instability of the ME at these extreme conditions [44–46]. Keeping SDS in the ME was important to accumulate the overall www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2015, 36, 2736–2744
2741
increasing amounts of SDS (Figs. 2A and B). This observation can be explained by the fact that with a higher concentration of SDS, a higher number of negatively charged micelles will get attracted to the cathode more easily, particularly when applying the reversed polarity mode. In turn, this will result into a short residence time within the capillary. 3.1.5 Optimum ME condition
negative charge which will make the FQ—micelle complex migrate to the cathode under the reversed polarity.
The overall optimum of the ME composition was 81.75% (w/w) of a 125 mM NaH2 PO4 buffer solution (pH = 2.75), 2.65% (w/w) SDS, 1.00% (w/w) n-octanol, 6.60% (w/w) n-butanol and 8.00% (w/w) IPA. This was derived from visual inspection of the respective surface response plots. The values were selected based on the multivariate examination of the response surface plots in order to strike a compromise in resolving all FQs within a reasonable analysis time and maintaining other separation parameters. A typical electropherogram of the separation of a mixture of the five FQs is depicted in Fig. 3.
3.1.3 Influence of isopropanol
3.2 Forced degradation studies
The percentage of IPA within the ME was checked at levels of 4.0, 6.0, 8.0, 10.0 and 12.0% (w/v). Increasing the ratio of IPA (⬎10% (w/w)) gradually prolonged the migration time of OFX (Fig. 2D), particularly when combined with high amounts of n-butanol and low concentrations of SDS. The plausible explanation has been given in Section 3.1.2. Adding an organic modifier does not only affect the selectivity but also the quality of the separation. This could be estimated from the peak efficiency and peak symmetry factors. Extended migration times were also associated with peak distortion and loss of peak symmetry as shown in Fig. 2B, depicting the multivariate influence of IPA and SDS on the asymmetry factor of OFX.
The study was performed according to the conditions given in Table 1. All FQs were stable under thermal stressing conditions which is in line with previous reports [19]. However, a significant degradation was observed to a variable extent (ranging from 90 to 70% of the API, depending on the individual FQ) at alkaline or acid hydrolysis, oxidative peroxide degradation and oxidative peroxide with catalyst degradation, respectively. The results of the forced degradations at various conditions are summarized in Table 1. Although most of the FQ impurities being described in the Ph. Eur. are of synthetic origin, the study went further to depict the “worst case” scenario for these products when undergoing degradation. The optimized method proved to be able to resolve the potential degradation products from each individual FQ and therefore substantiates the claim of a stability-indicating generic method. An overlay of the electropherograms from five test solutions stressed under oxidative peroxide condition (cf. Table 1) is presented in Fig. 4A. Many compounds could be well resolved (MFX: eight impurity peaks, CFX: 11 impurity peaks, NFX: 13 impurity peaks, LFX and OFX: six impurity peaks). Forced degradation under acidic condition yielded relatively
Figure 3. A typical electropherogram of a mixture of the five FQ antibiotics acquired under the optimized MEEKC conditions (resolution CFX/NFX = 2.75).
3.1.4 Influence of SDS The amount of SDS was varied and five concentrations were investigated: 1.35, 2.0, 2.65, 3.3 and 3.95% (w/w). A better selectivity between CFX and NFX was observed when the amount of SDS in the ME was increased up to about 2.65% (w/w), whereas OFX asymmetry decreased constantly with
Table 1. Summary of degradation profiles of the forced degradation studies at various conditions
Condition
Alkaline hydrolysis: 0.1 M NaOH Acid hydrolysis: 0.1 M H2 SO4 Oxidative without catalyst: 10% H2 O2 Oxidative with catalyst: 3% H2 O2 + 1% FeCl3 Thermal degradation (in H2 O)
Treatment
Kept for 24 h, then refluxed (100°C, 60 min) Kept for 24 h, then refluxed (100°C, 60 min) Refluxed (100°C, 60 min) Refluxed (100°C, 30 min) Refluxed (100°C, 24 h)
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Number of peaks observed MFX
CFX
NFX
OFX
LFX
Mixture
4 14 – 8 –
3 3 1 11 –
4 6 2 13 –
3 6 3 6 –
2 7 3 6 –
Not determined Not determined Not determined 23 –
www.electrophoresis-journal.com
2742
E. Kaale et al.
Electrophoresis 2015, 36, 2736–2744
3.3 Method validation 3.3.1 Specificity and peak identification
Figure 4. Overlaid electropherograms of the five FQs after oxidative (A) and acidic (B) forced degradation.
few signals and is represented as overlaid electropherograms in Fig. 4B. The bumpy baseline might have resulted from an unknown compound (e.g. an excipient) which might be present in the proprietary product, or the acidic solution, respectively. However, this does not interfere with the peaks of interest.
It was observed after spiking the FQs with the respective related substances (cf. Fig. 1) that all commercially available impurities could be resolved well from the main component. In addition, evaluation of the peak purity confirmed that all separated peaks were due to one compound, each. In addition to the known impurities, several unknown substances could also be separated. Figure 5 presents the overlaid electropherograms of CFX spiked with its impurity A, NFX spiked with its impurity A and OFX spiked with its impurities A and E. These impurities were commercially available as pure compounds and thus, the peaks could be assigned. According to the certificate of analysis which was delivered together with “Ciprofloxacin for peak identification CRS” it was indicated that the HPLC method from the Ph. Eur. [24] is able to resolve four impurities (i.e. B, C, D and E) from the API. Of note, our proposed method was able to separate a total of eight impurities plus the main substance (Supporting Information Fig. S2A), thus offering a striking advantage in the sense that only one analysis run has to be performed to assess this drug. This is not the case in the methods which have been added to the Ph. Eur. and the International Pharmacopoeia (Supporting Information Figs. S1A and B) [8, 10]. In analogy, additional unknown signals could be observed in the reference standards of OFX impurity E, NFX impurity A and NFX itself (Supporting Information Fig. S3).
3.3.2 Linearity and accuracy Evaluation of linearity for CFX, NFX, MFX, OFX and LFX was carried out via plotting the corrected peak area ratios against an internal standard in the concentration range 0.04– 0.48 mg/ml. For all five FQs, a good linear relationship was
Figure 5. Overlaid electropherograms of CFX and NFX spiked with their respective impurity A as well as OFX spiked with its impurities A and E using the optimized MEEKC method.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2015, 36, 2736–2744
4 Concluding remarks
Table 2. Method accuracy summary for the five FQs at three levels (lower, nominal and upper)
MFX
CFX
NFX
LFX
OFX
Level (%)
Added (mg/ml)
Observed (mg/ml)
Accuracy (%)
sd
rsd (%)
C1–80 C2–100 C3–120 C1–80 C2–100 C3–120 C1–80 C2–100 C3–120 C1–80 C2–100 C3–120 C1–80 C2–100 C3–120
0.32 0.40 0.48 0.31 0.40 0.48 0.32 0.40 0.47 0.32 0.41 0.48 0.32 0.39 0.47
0.32 0.39 0.46 0.30 0.39 0.47 0.32 0.39 0.47 0.31 0.40 0.47 0.32 0.39 0.47
98.47 98.00 96.72 97.45 97.77 99.48 100.58 97.83 99.36 99.36 98.36 97.60 100.60 100.84 100.08
1.78 1.64 1.93 2.92 1.75 2.66 2.43 2.22 2.78 2.78 2.18 1.25 1.73 1.17 1.66
1.81 1.67 1.99 3.00 1.79 2.68 2.41 2.27 2.80 2.80 0.02 0.01 1.71 1.16 1.66
Table 3. Repeatability and intermediate precision
Repeatability (n = 6)
MFX CFX NFX LFX OFX
Intermediate precision (n = 12)
Mean corrected peak area/IS
sd
rsd (%)
Mean corrected peak area/IS
sd
rsd (%)
0.53 0.93 0.55 1.61 1.68
0.01 0.03 0.01 0.03 0.04
2.32 3.15 2.45 1.64 2.27
0.53 0.55 1.61 0.93 0.99
0.01 0.01 0.03 0.03 0.02
2.23 2.67 2.04 3.15 1.54
obtained. The regression data were y = 0.008x – 0.002 (R = 0.994) for MFX, y = 0.017x + 0.002 (R2 = 0.997) for CFX, y = 0.016x – 0.0001 (R2 = 0.997 for NFX), y = 0.009x + 0.020 (R2 = 0.989) for OFX and y = 0.009x + 0.020 (R2 = 0.989) for LFX, respectively. Alternatively, the ANOVA T- and F-test at a desired significance of 0.05 gave p-values below 0.001, for all five substances. H0 :1 is rejected, implying that a linear relation does exist between concentration and the responses. The percentage accuracies for the determination assay of CFX, NFX, OFX and LFX for the proposed generic MEEKC method are summarized in Table 2. 2
3.3.3 Precision The repeatability and intermediate precision data for the selected FQs are presented in Table 3. The values were calculated by using the mean corrected peak area ratio to the internal standard, expressed in terms of %rsd and were found to range from about 1 to 3%, suggesting the method has an adequate level of precision for the assay of the target drugs. C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
2743
A generic stability-indicating MEEKC method has been developed and validated. The protocol is able to separate five FQs in a mixture, or the respective impurities from the individual FQ within 20 min in a single run. The proposed MEEKC method is linear, simple, precise and accurate. It was also possible to separate impurities A, B, C, D and E of CFX and impurities A and E of OFX within one run which is not possible when using the official Ph. Eur. method. The method is representing an “all-in-one” concept which we anticipated before, meaning that for a testing laboratory the amount of required chemicals and supplies can be minimized [47]. This is a striking advantage in a resourceconstrained setting. The method was furthermore used to study the degradation behaviour of the five FQs in different media indicating that it is suitable for evaluating the stability of a drug sample. The authors gratefully thank the Alexander von HumboldtFoundation (Bonn, Germany) for awarding the Georg Forster Research Award for Senior Research Fellows to Dr. Eliangiringa Kaale. The Muhimbili University of Health and Allied Sciences is highly acknowledged for enabling the sabbatical leave of Dr. Eliangiringa Kaale. The authors declare no conflict of interest.
5 References [1] Anderson, R., Groundwater, P. W., Todd, A., Worsley, A. J., Antibacterial Agents: Chemistry, Mode of Action, Mechanisms of Resistance and Clinical Applications, Wiley, Chichester 2012. [2] Hooper, D. C., Biochim. Biophys. Acta 1998, 1400, 45–61. [3] Wolfson, J. S., Hooper, D. C., Clin. Microbiol. Rev. 1989, 2, 378–424. [4] Borcherding, S. M., Stevens, R., Nicholas, R. A., Corley, C. R., Self, T., J. Fam. Pract. 1996, 42, 69–78. [5] Huang, J., Kaul, G., Cai, C., Chatlapalli, R., HernandezAbad, P., Ghosh, K., Nagi, A., Int. J. Pharm. 2009, 382, 23–32. [6] Yu, L. X., Pharm. Res. 2008, 25, 781–791. [7] Yu, L. X., Amidon, G., Khan, M. A., Hoag, S. W., Polli, J., Raju, G. K., Woodcock, J., AAPS J. 2014, 16, 771–783. [8] European Pharmacopoeia, 8th Edition, Monograph “Ciprofloxacin hydrochloride” No. 04/2015:0888, European Directorate for the Quality of Medicines & HealthCare, Strasbourg, France 2014. [9] European Pharmacopoeia, 8th Edition, Monograph “Norfloxacin” No. 04/2011:1248, European Directorate for the Quality of Medicines & HealthCare, Strasbourg, France 2014. [10] The International Pharmacopoeia, 4th Edition, Health Organization, World Health Organization, Geneva, Switzerland 2010.
www.electrophoresis-journal.com
2744
E. Kaale et al.
[11] Miseljic, B., Popovic, G., Agbaba, D., Markovic, S., Simonovska, B., Vovk, I., J. AOAC Int. 2008, 91, 332– 338. [12] Cazedey, E. C. L., Salgado, H. R. N., Adv. Anal. Chem. 2012, 2, 74–79. [13] Natraj, K. S., Suvarna, Y., Prasanti, G., Saikumar, S. V., Int. Res. J. Pharm. 2013, 4, 178–181. [14] Argekar, A. P., Sawant, J. G., J. Planar Chromatogr. 1999, 12, 202–206. [15] Novakovic, J., Nesmerak, K., Nova, H., Filka, K., J. Pharm. Biomed. Anal. 2001, 25, 957–964. [16] Singh, B. K., Parwate, D. V., Srivastava, S., Shukla, S. K., Pharma. Chem. 2010, 2, 178–188. [17] Nyamweru, B. C., Kaale, E., Manyanga, V. P., Chambuso, M., Layloff, T., J. Planar Chromatogr. 2013, 26, 370– 374.
Electrophoresis 2015, 36, 2736–2744
[28] Li, L., Xia, Z., Yang, F., Chen, H., Zhang, Y., J. Sep. Sci. 2012, 35, 2101–2107. [29] Cruz, L. A., Hall, R., J. Pharm. Biomed. Anal. 2005, 38, 8–13. [30] Novakovic, J., Nesmerak, K., Nova, H., Filka, K., J. Pharm. Biomed. Anal. 2001, 25, 957–964. [31] Krzek, J., Hubicka, U., Szczepanczyk, J., J. AOAC Int. 2005, 88, 1530–1536. [32] Schulman, J. H., Stoeckenius, W., Prince, L. M., J. Phys. Chem. 1959, 63, 1677–1680. [33] McEvoy, E., Marsh, A., Altria, K., Donegan, S., Power, J., Electrophoresis 2007, 28, 193–207. [34] Ryan, R., Donegan, S., Power, J., McEvoy, E., Altria, K., Electrophoresis 2009, 30, 65–82. [35] Ryan, R., McEvoy, E., Donegan, S., Power, J., Altria, K., Electrophoresis 2011, 32, 184–201.
[18] Aksoy, B., Kucukguzel, I., Rollas, S., Chromatographia 2007, 66, S57-S63.
[36] Marsh, A., Clark, B., Broderick, M., Power, J., Donegan, S., Altria, K., Electrophoresis 2004, 25, 3970–3980.
[19] Lalitha Devi, M., Chandrasekhar, K. B., J. Pharm. Biomed. Anal. 2009, 50, 710–717.
[37] De Lu, J., Yuan, W., Fu, X. Y., Chin. Chem. Lett. 2001, 12, 155–156.
[20] Mehta, J., Pancholi, Y., Patel, V., Kshatri, N., Vyas, N., Int. J. PharmTech Res. 2010, 2, 1932–1942.
[38] Schwuger, M.-J., Stickdorn, K., Schomaecker, R., Chem. Rev. 1995, 95, 849–864.
[21] Reddy, B. V., Kumar, A. P., Reddy, G. V. R., Sahai, M., Sreeramulu, J., Park, J. H., Anal. Lett. 2010, 43, 2653–2662.
[39] Lamalle, C., Djang’Eing’A, M. R., Debrus, B., Lebrun, P., Crommen, J., Hubert, P., Servais, A.-C., Fillet, M., Electrophoresis 2012, 33, 1669–1678.
[22] Shervington, L. A., Abba, M., Hussain, B., Donnelly, J., J. Pharm. Biomed. Anal. 2005, 39, 769–775.
[40] The United States Pharmacopoeia USP 37 - NF 32, The United States Pharmacopoeial Convention, Rockville, MD, USA 2014.
[23] Faria, A. F., de Souza, M. V. N., de Oliveira, M. A. L., J. Braz. Chem. Soc. 2008, 19, 389–396. [24] European Pharmacopoeia, 8th Edition, Monograph “Ofloxacin” No. 01/2011:1455, European Directorate for the Quality of Medicines & HealthCare, Strasbourg, France 2014. [25] European Pharmacopoeia, 8th Edition, Monograph “Ciprofloxacin” No. 04/2011:1089, European Directorate for the Quality of Medicines & HealthCare, Strasbourg, France 2014.
[41] Guideline Q2(R1): Validation of Analytical Procedures: Text and Methodology, International Conference on Harmonisation, 2006, http://www.ich.org ´ S., Horvat, A. J. M., Mutavdˇzic´ Pavlovic, ´ D., [42] Babic, ˇ Kastelan-Macan, M., TrAC Trends Anal. Chem. 2007, 26, 1043–1061. [43] Baeuml, F., Welsch, T., J. Chromatogr. A 2002, 961, 35–44. [44] Chang, C. W., Chen, Y. C., Liu, C. Y., Electrophoresis 2014, 35, 2901–2906.
[26] The United States Pharmacopoeia USP 37 – NF 32, Monographs “Ciprofloxacin” and “Ciprofloxacin hydrochloride”, The United States Pharmacopoeial Convention, Rockville, MD, USA 2014.
[45] Pomponio, R., Gotti, R., Luppi, B., Cavrini, V., Electrophoresis 2003, 24, 1658–1667.
[27] Al Azzam, K. M., Saad, B., Adnan, R., Aboul-Enein, H. Y., Anal. Chim. Acta 2010, 674, 249–255.
[47] Hoellein, L., Holzgrabe, U., J. Pharm. Biomed. Anal. 2014, 98, 434–445.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[46] Marsh, A., Clark, B. J., Altria, K. D., J. Sep. Sci. 2005, 28, 2023–2032.
www.electrophoresis-journal.com
¨ of Wurzburg, Institute of Pharmacy and Food ¨ Chemistry, Wurzburg, Germany 2 School of Pharmacy, Muhimbili University of Health and Allied Sciences, Dar Es Salaam, Tanzania
Received January 19, 2015 Revised April 21, 2015 Accepted April 21, 2015
Electrophoresis 2015, 36, 2736–2744
Research Article
Development and validation of a generic stability-indicating MEEKC method for five fluoroquinolone antibiotics In this paper, a reliable stability-indicating generic MEEKC method for the analysis of five commonly used fluoroquinolones (FQs) has been developed and optimized by a central composite circumscribed experimental design. The separation was carried out using a fused silica capillary (60.2 cm total length) and a microemulsion (ME) composed of 81.75% (w/w) of a 125 mM NaH2 PO4 solution having a pH of 2.75, 2.65% (w/w) SDS, 1.00% (w/w) n-octanol, 6.60% (w/w) n-butanol and 8.00% (w/w) 2-propanol. A voltage of 28 kV was applied in a reverse polarity mode. A linear relationship was established from 0.04 to 0.48 mg/ml with R2 values higher than 0.98 for all five FQs. Both repeatability and intermediate precision were less than 3% and accuracy ranging from 97 to 100%. Of note, ciprofloxacin impurity A and ofloxacin impurity A could be separated from the respective drug substance in a single run which cannot be achieved using the official HPLC method from the European Pharmacopoeia. Forced degradation of all FQs under heat, in acidic and alkaline medium, in the presence of oxidizing agents and under neutral hydrolysis conditions was investigated. The highest yield of degradation products was observed using oxidative hydrogen peroxide. Hence, the proposed MEEKC method can be used for the quantitative determination of the five FQs and their potential impurities within a total runtime of 20 min. Keywords: Fluoroquinolone antibiotics / Impurity profiling / MEEKC / Stability-indicating method DOI 10.1002/elps.201500025
Additional supporting information may be found in the online version of this article at the publisher’s web-site
1 Introduction Fluoroquinolones (FQs) belong to an important group of antibiotics which are commonly used in human and veterinary medicine for the treatment of a wide variety of infections. Their activity is based on the interference with various topoisomerases [1]. This makes them broad-range antibiotics
Correspondence: Professor Eliangiringa Kaale, School of Pharmacy, Muhimbili University of Health and Allied Sciences, Dar Es Salaam, Tanzania, P.O. Box 65545/11103, Upanga West, Dar Es Salaam, Tanzania. E-mail: [email protected]
Abbreviations: API, active pharmaceutical ingredient; CCD, central composite design; CFX, ciprofloxacin; FQ, fluoroquinolone; IPA, isopropanol; LFX, levofloxacin; LoFX, lomefloxacin; ME, microemulsion; MFX, moxifloxacin; NFX, norfloxacin; OFX, ofloxacin; PDA, Photodiode Array Detector; Ph. Eur., European Pharmacopoeia; SC, sodium cholate; TEA, trietylamine
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
which are effective against gram-positive and -negative bacteria [2–4]. Usually the quality of commercially available finished pharmaceutical products is determined by several critical quality attributes [5]: (i) identity, (ii) assay, (iii) disintegration, (iv) dissolution and (v) impurity profile [6,7]. Developing separation methods which are able to determine the related substances of a compound is a major issue during and after drug development. Routinely applying such methods still remains a challenge for quality control laboratories particularly in developing countries with restricted resources. In many cases, compendial methods are cumbersome or tedious and often require the application of more than one technology in order to be able to resolve all related substances [8–10]. However, the availability of suitable analytical methods which are able to separate an active pharmaceutical ingredient (API) from all its potential impurities is paramount during the quality assessment of pharmaceutical products and the detection of counterfeits [11]. For the recent ten years several analytical protocols for the determination of commonly available FQs have
www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2015, 36, 2736–2744
2737
Figure 1. Chemical structures of the investigated FQs and the internal standard lomefloxacin.
been published such as UV spectroscopy for quantitative analysis [12, 13], or TLC and high-performance thin layer chromatography (HPTLC) methods [14–17] as well as stability-indicating HPLC methods [18–22] and CZE, respectively [23]. The monographs which have been added to the European Pharmacopoeia (Ph. Eur.) describe LC tests for the control of related substances of the most important FQs. Nevertheless, it is not always possible to resolve all potential impurities using one technology only, for example in the case of ciprofloxacin (CFX), norfloxacin (NFX) and ofloxacin (OFX). The respective monographs which can be found in the Ph. Eur. and the United States Pharmacopoeia (USP) describe the combination of TLC and LC in order to determine the related substances of the compounds [8–10, 24–26]. Enantioselective methods for OFX [27, 28] and moxifloxacin (MFX) [29] have also been published; all of them apply CZE in combination with a chiral selector in order to resolve the respective enantiomers. On the other hand, simpler methods for determining known and/or unknown impurities are available for the analysis of CFX [30, 31]. Microemulsions (MEs) have been described for the first time in 1959 [32] and prove to be a potential methodology for separating an array of compounds with diverse physicochemical properties [33, 34]. They are isotropically clear solutions consisting of a lipophilic component (e.g. octane) and water, and are stabilized by adding a surfactant or a co-surfactant whose presence reduces the interfacial tension to almost zero [35, 36]. Today ME systems are widely applied in CE in a mode called MEEKC [37] which utilizes MEs as the separation medium [38]. Recently, a generic MEEKC method was successfully developed and applied for the analysis of 15 different antimalarials [39]. This is a great benefit for the practical work particularly in resource-constrained settings because a testing laboratory would only need one inventory of supplies catering for all 15 drug products instead of having 15 inventories for each individual API.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
To the best of our knowledge there is no existing generic MEEKC method for the separation of the commonly used FQs CFX, NFX, MFX, levofloxacin (LFX) and OFX including their potential related substances (Fig. 1 and Supporting Information Fig. S1). Our study aimed at developing and validating a generic MEEKC method for the separation and quantitative determination of these five mostly used FQs by applying an experimental design space approach to optimize the critical separation parameters. In addition, the purity of the single compounds can be analysed as the method is also able to resolve the respective related substances from the main signal when individual runs are performed for each API. During method development commercially available and known impurities which are described in the respective monographs of the Ph. Eur. were considered. An experimental design space approach was applied to optimize the critical MEEKC parameters and to (i) enable the quantitative determination of the compounds, (ii) to detect counterfeits with regard to the exchange of the API and (iii) to separate the respective impurities from the individual APIs and thus determining the purity of the compound.
2 Materials and methods 2.1 Apparatus and electrophoretic conditions A Beckman P/ACE MDQ capillary electrophoresis instrument was applied (Beckman Coulter GmbH, Krefeld, Germany), equipped with a capillary cartridge cooling system and a photodiode array detector (PDA). Data were acquired and processed using the Beckman 32 Karat software (version 4.01). The capillary was 60.2 cm long (50 cm effective length, internal diameter 50 m) and supplied from BGB Analytik (Schloßb¨ockelheim, Germany). The applied voltage was 28 kV (0.17 min ramp time) using reverse polarity. The capillary was thermostated at 25°C and all samples were injected applying a pressure of 0.5 psi for 10 s. The PDA
www.electrophoresis-journal.com
2738
E. Kaale et al.
was set at 295 nm using a data sampling rate of 4 Hz, a bandwidth of 10 nm, and normal filtering, respectively. New capillaries were preconditioned at 40°C by rinsing them with 0.1 M NaOH (10 min) – purified water (5 min) – 0.1 M H3 PO4 (10 min), and purified water (5 min) at a pressure of 40 psi. Between two runs the capillary was flushed using the following reagents at 40 psi and 25°C: 0.1 M NaOH for 5 min, methanol for 3 min, 0.1 M H3 PO4 for 3 min, purified water for 3 min and running buffer for 5 min. The pH values were determined using a calibrated glass electrode from Radiometer Analytical (Villeurbanne, France). All solutions were passed through 0.22 m syringe filters prior to use. The optimization by experimental design and multivariate analysis was carried out using the Modde 4.0 software (Umetrics, Umea, ˚ Sweden).
2.2 Chemicals and materials SDS, Tris-phosphate, sodium cholate (SC), H3 PO4 (85% v/v), NaOH, anhydrous NaH2 PO4 , FeCl3 , triethylamine (TEA), H2 O2 , n-butanol, n-octanol as well as CFX, lomefloxacin (LoFX) and LFX primary standards were obtained from Sigma-Aldrich Chemie (Taufkirchen, Germany). Isopropanol (IPA) and methanol were purchased from Fisher Scientific (Loughborough, UK). OFX and NFX primary standards were supplied through Alfa Aesar GmbH & Co KG (Karlsruhe, Germany). MFX hydrochloride, CFX impurity A, NFX impurities A and E, OFX impurities A and E as well as “Ciprofloxacin hydrochloride for peak identification CRS” and “Norfloxacin for peak identification CRS” were obtained from the EDQM (Strasbourg, France). Two CFX commercial samples manufactured by Riva Pharma, Cairo, Egypt, were bought in Dar Es Salaam, Tanzania. High purity water was prepared by a Milli-Q water purification system (Merck Millipore, Billerica, USA).
2.3 Preparation of the microemulsion The background electrolyte (125 mM NaH2 PO4 ) was prepared as follows: 7.5 g of anhydrous NaH2 PO4 were dissolved in about 400 ml Millipore water, the pH of the solution was adjusted to 2.75 with 1.0 M H3 PO4 and the volume was made up to 500 ml with water. The standard o/w ME was prepared by mixing 0.5 g n-octanol, 3.3 g n-butanol, 1.33 g SDS, 4.0 g of IPA and 40.88 g of the background electrolyte solution. The mixture was either stirred for 1 h or sonicated for 30 min to obtain an optically transparent ME.
Electrophoresis 2015, 36, 2736–2744
MeOH (85:15 v/v). The solution was sonicated for 10 min to facilitate dissolution. Diluent B: 25 mM H3 PO4 and MeOH were mixed in the ratio 85:15 (v/v). For preparing the standard solutions, 50.0 mg of the respective FQ were individually dissolved in 50.0 ml of diluent A, whereas for the market sample stock solutions portions equivalent to 50.0 mg of the respective FQ were weighed from crushed tablets and dissolved in 50.0 ml of diluent A. Standard and sample working solutions were prepared by diluting 10.0 ml aliquots of the respective stock solution to a final volume of 25.0 ml using the same diluent (0.4 mg/ml). For peak identification purposes, portions equivalent to 50.0 mg of the respective FQ were weighed from crushed tablets and dissolved in 50.0 ml of diluent B, and in the case of CFX and NFX, 10.0 mg of “Ciprofloxacin hydrochloride for peak identification CRS” and 10.0 mg of “Norfloxacin for peak identification CRS”, respectively, were individually dissolved in 20.0 ml of diluent B. Electropherograms obtained from these solutions were used to determine the purity of the commercial CFX and NFX tablets by comparison of the individual peak patterns. For the forced degradation studies the test substances were individually dissolved in 0.1 M NaOH, 0.1 M H2 SO4 and a 3% solution of H2 O2 with and without the addition of 1% FeCl3 , respectively, to obtain solutions of 0.4 mg/ml. The alkaline and acidic solutions were kept for 24 h and then refluxed at 100°C for 60 min whereas the peroxide solutions were refluxed at 100°C for 30 min. A thermal degradation test was carried out by dissolving the five APIs in a mixture of 25 mM H3 PO4 and MeOH (85:15 v/v) and refluxing at 100°C for 24 h.
2.5 Method development and optimization by design of experiments (DoE) 50.0 mg of each FQ were dissolved in 50.0 ml of diluent B, and 10.0 ml of this solution were diluted to a final volume of 25.0 ml using the same solvent (0.4 mg/ml). Compound separation and optimization of the method was carried out using this model solution. The separation of the individual FQs from their impurities was investigated by preparing solutions containing the API and their respective impurities (i.e. CFX and impurities A, B, C, D and E; NFX and impurities A, E, H and K; MFX, and OFX and impurities A and E; see Fig 1). The most important factors affecting the separation of the compounds were selected for method optimization by applying a central composite response surface modelling experiment, and from screening trials a total set of 27 experiments was derived by application of a central composite design (CCD, Supporting Information Table S1).
2.4 Sample solution preparation All solutions, unless stated otherwise, were prepared using two different diluents prepared as follows: Diluent A: 50.0 mg of LoFX (internal standard) were dissolved in 500.0 ml of a mixture of 25 mM phosphoric acid and C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
2.6 Method validation Validation was carried out with reference to the methodologies described in the guideline Q2(R1) “Validation of www.electrophoresis-journal.com
Electrophoresis 2015, 36, 2736–2744
Analytical Procedures” of the International Conference on Harmonisation (ICH)., and in the USP [40, 41].
2.6.1 Specificity Spiking FQ samples with available impurity standards (cf. Section 2.5) were applied for compound identification and assessment of selectivity. If impurities described in the Ph. Eur. were not commercially available, electropherograms of the samples were compared to those obtained with standard solutions and the forced degradation studies. Since the excipients used for the formulation of the proprietary products were not known the comparison with a blank run (i.e. a placebo mixture) was not possible.
2.6.2 Linearity and accuracy A stock solution was prepared by weighing amounts equivalent to 50.0 mg of each FQ and individually dissolving them in 50.0 ml of diluent A. Serial dilutions using the same solvent were made to obtain five levels (10, 40, 80, 100 and 120%) by standard addition method. Independent controls were prepared at levels of 80, 100 and 120%. The calibrator (i.e. S1-S5) and control (i.e. C1-C3) solutions were randomized on a sequence as S1-C1-C2-S2-C3-S3-C1-C2-C3-S4-C1-C2-C3-S5 in order to minimize errors due to possible variations of the sample or the ME over time. Calibration curves were constructed by plotting the peak responses vs. the respective concentration. Accuracy was determined by evaluating the mean recovery of the analyte at the concentration levels of 80, 100 and 120% from a spiked sample, i.e. aqueous solutions of the compounds which were extracted from tablets. An ANOVA test was performed in order to investigate the fitness of the calibration model.
2.6.3 Precision Repeatability and intermediate precision were determined by independently preparing six replicate working solutions at concentrations corresponding to a level of 100% as described in Section 2.4. Intermediate precision was assessed using a different technical equipment and a different laboratory day.
3 Results and discussion 3.1 Method development: preliminary trials and optimization of the MEEKC separation conditions by DoE A total of approach n-octanol, respective
four variables was investigated in a multivariate by varying the percentage concentrations of n-butanol, IPA and SDS within the ME (see section). Four the most influential variables were
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
CE and CEC
2739
selected and investigated using CCD at five levels in the following concentration ranges: 1.35–3.95% (w/w) SDS, 4.6–8.6% (w/w) n-butanol, 0.6–1.2% (w/w) n-octanol and 4–12% (w/w) n-octanol, respectively. Table S1 (Supporting Information) gives a summary matrix for the CCD design used in the multivariate analysis. The resolution Rs corresponding to the critical peak pair CFX-NFX, the migration time, the number of theoretical plates and the asymmetry factor of the last migrating peak OFX were selected as response variables determining the separation quality. It was assumed that if the generic method would be able to resolve the five FQs, it could also resolve the related substances from each individual API. This was confirmed through spiking throughout the method development. Optimal separation conditions were derived from preliminary trials where Tris-phosphate, NaH2 PO4 and TEA were used at different concentrations (20–200 mM) and pH values (1.85– 3.50). The results indicated that with TEA and Tris-phosphate buffer no separation between CFX, NFX and OFX could be achieved. However, only MFX was resolved from all other signals. Using a 125 mM NaH2 PO4 buffer at pH 2.75, at least two peaks were baseline separated while CFX and NFX comigrated. As the apparent pH of the ME system was 2.95 and the pKa values of the basic nitrogen atoms ranged between 6 and 9, the amino groups were fully protonated [42]. In reversed polarity mode, one would not expect them to move to the anode considering that the EOF is almost suppressed at the low pH value. With SDS being present in the ME the cationic FQs are incorporated into micelles through ionic interaction, which probably get attracted to the anode due to the overall negatively charged complex. In addition, different types of surfactants (SC, Brij-35, SC+Brij-35, SC+Triton X-100, SDS+SC) were investigated. Of note, neither using an individual surfactant alone nor combining them could deliver better results compared to the application of SDS in the concentration range between 2 and 3.5%. From this experiment, SDS was chosen in all further experiments. In order to investigate the influences of temperature on the separation quality the capillary temperature was varied from 20 to 30°C at an interval of 5°C. At lower temperature the OFX migration time was longer than 40 min, however its peak symmetry and the separation from the other compounds were improved at 25°C. At 30°C the current became too high. Subsequently, the capillary temperature was fixed at 25°C and the influence of various oils (i.e. n-octane, n-octanol, n-heptane and ethyl acetate) was evaluated. Using n-octanol at concentrations between 0.5 and 1.0% yielded an improved separation between CFX and NFX, however the number of theoretical plates decreased in this range. The influence of the commonly used co-surfactant, n-butanol, was also investigated between 4 and 8% (w/w), and both resolution and number of the theoretical plates could be raised to 600 000 for the last migrating peak OFX. Adding small organic modifiers to facilitate solubility and partitioning was also investigated, and 4 to 15% (w/w) of ACN, MeOH as well as IPA, respectively, were added. Using ACN resulted into frequent capillary breaks as it was dissolving the www.electrophoresis-journal.com
2740
E. Kaale et al.
Electrophoresis 2015, 36, 2736–2744
Figure 2. Response surface plots displaying the multifactorial influence of variables on the separation parameters.
polyimide coating within the first two or three runs which is consistent with observations previously published by Baeuml and Welsch [43]. The results indicated that the resolution between the critical peak pair CFX-NFX reached a maximum using 8–10% (w/w) of IPA, whereas at higher amounts (15% (w/w)) it decreased and the analysis time went up to 50 min. Using MeOH instead of ACN and IPA could not improve the separation of the compounds at all. Thus, IPA was selected. The molarity of the phosphate buffer, its pH value and the capillary temperature were fixed at 125 mM, 2.75 and 25°C, respectively. 3.1.1 Influence of n-octanol From the initial trials it was observed that CFX and NFX were migrating closely. This might be due to their similar chemical structures, as the only difference is a cyclopropyl instead of the ethyl substituent on the piperazine ring in position 7 (cf. Fig. 1). However, with an increasing octanol concentration, the resolution between CFX and NFX gradually improved up to 3.2 at 1% (w/w) n-octanol; with higher octanol concentrations no better resolution could be achieved, as it is illustrated in Fig. 2A. The influences were studied at five levels, i.e. 0.6, 0.8, 1.0, 1.2 and 1.4% (w/w). C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The combined multivariate interaction effect with IPA has also shown to have a significantly positive effect in resolving this peak pair (results not shown). n-Octanol had a relatively weak effect on both migration times and asymmetry factors of OFX.
3.1.2 Influence of n-butanol The influence of n-butanol in the ME was studied at 4.6, 5.6, 6.6, 7.6 and 8.6% (w/w), respectively, and a positive effect of using it in the ME could be observed even at low concentrations. A steadily increase of the number of theoretical plates was found up to a percentage of 6.6% (w/w). A further increase resulted into a gradually decrease of the peak efficiency, in a prolonged migration time and a large asymmetry of the OFX peak. The multivariable response surface plot is depicted in Fig. 2D. A significant negative multivariate interaction effect was found with IPA, thus combining a high amount of n-butanol and IPA as well as lowering the concentration of SDS resulted into a longer analysis time. Two experiments with this combination were removed from the model because of the instability of the ME at these extreme conditions [44–46]. Keeping SDS in the ME was important to accumulate the overall www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2015, 36, 2736–2744
2741
increasing amounts of SDS (Figs. 2A and B). This observation can be explained by the fact that with a higher concentration of SDS, a higher number of negatively charged micelles will get attracted to the cathode more easily, particularly when applying the reversed polarity mode. In turn, this will result into a short residence time within the capillary. 3.1.5 Optimum ME condition
negative charge which will make the FQ—micelle complex migrate to the cathode under the reversed polarity.
The overall optimum of the ME composition was 81.75% (w/w) of a 125 mM NaH2 PO4 buffer solution (pH = 2.75), 2.65% (w/w) SDS, 1.00% (w/w) n-octanol, 6.60% (w/w) n-butanol and 8.00% (w/w) IPA. This was derived from visual inspection of the respective surface response plots. The values were selected based on the multivariate examination of the response surface plots in order to strike a compromise in resolving all FQs within a reasonable analysis time and maintaining other separation parameters. A typical electropherogram of the separation of a mixture of the five FQs is depicted in Fig. 3.
3.1.3 Influence of isopropanol
3.2 Forced degradation studies
The percentage of IPA within the ME was checked at levels of 4.0, 6.0, 8.0, 10.0 and 12.0% (w/v). Increasing the ratio of IPA (⬎10% (w/w)) gradually prolonged the migration time of OFX (Fig. 2D), particularly when combined with high amounts of n-butanol and low concentrations of SDS. The plausible explanation has been given in Section 3.1.2. Adding an organic modifier does not only affect the selectivity but also the quality of the separation. This could be estimated from the peak efficiency and peak symmetry factors. Extended migration times were also associated with peak distortion and loss of peak symmetry as shown in Fig. 2B, depicting the multivariate influence of IPA and SDS on the asymmetry factor of OFX.
The study was performed according to the conditions given in Table 1. All FQs were stable under thermal stressing conditions which is in line with previous reports [19]. However, a significant degradation was observed to a variable extent (ranging from 90 to 70% of the API, depending on the individual FQ) at alkaline or acid hydrolysis, oxidative peroxide degradation and oxidative peroxide with catalyst degradation, respectively. The results of the forced degradations at various conditions are summarized in Table 1. Although most of the FQ impurities being described in the Ph. Eur. are of synthetic origin, the study went further to depict the “worst case” scenario for these products when undergoing degradation. The optimized method proved to be able to resolve the potential degradation products from each individual FQ and therefore substantiates the claim of a stability-indicating generic method. An overlay of the electropherograms from five test solutions stressed under oxidative peroxide condition (cf. Table 1) is presented in Fig. 4A. Many compounds could be well resolved (MFX: eight impurity peaks, CFX: 11 impurity peaks, NFX: 13 impurity peaks, LFX and OFX: six impurity peaks). Forced degradation under acidic condition yielded relatively
Figure 3. A typical electropherogram of a mixture of the five FQ antibiotics acquired under the optimized MEEKC conditions (resolution CFX/NFX = 2.75).
3.1.4 Influence of SDS The amount of SDS was varied and five concentrations were investigated: 1.35, 2.0, 2.65, 3.3 and 3.95% (w/w). A better selectivity between CFX and NFX was observed when the amount of SDS in the ME was increased up to about 2.65% (w/w), whereas OFX asymmetry decreased constantly with
Table 1. Summary of degradation profiles of the forced degradation studies at various conditions
Condition
Alkaline hydrolysis: 0.1 M NaOH Acid hydrolysis: 0.1 M H2 SO4 Oxidative without catalyst: 10% H2 O2 Oxidative with catalyst: 3% H2 O2 + 1% FeCl3 Thermal degradation (in H2 O)
Treatment
Kept for 24 h, then refluxed (100°C, 60 min) Kept for 24 h, then refluxed (100°C, 60 min) Refluxed (100°C, 60 min) Refluxed (100°C, 30 min) Refluxed (100°C, 24 h)
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Number of peaks observed MFX
CFX
NFX
OFX
LFX
Mixture
4 14 – 8 –
3 3 1 11 –
4 6 2 13 –
3 6 3 6 –
2 7 3 6 –
Not determined Not determined Not determined 23 –
www.electrophoresis-journal.com
2742
E. Kaale et al.
Electrophoresis 2015, 36, 2736–2744
3.3 Method validation 3.3.1 Specificity and peak identification
Figure 4. Overlaid electropherograms of the five FQs after oxidative (A) and acidic (B) forced degradation.
few signals and is represented as overlaid electropherograms in Fig. 4B. The bumpy baseline might have resulted from an unknown compound (e.g. an excipient) which might be present in the proprietary product, or the acidic solution, respectively. However, this does not interfere with the peaks of interest.
It was observed after spiking the FQs with the respective related substances (cf. Fig. 1) that all commercially available impurities could be resolved well from the main component. In addition, evaluation of the peak purity confirmed that all separated peaks were due to one compound, each. In addition to the known impurities, several unknown substances could also be separated. Figure 5 presents the overlaid electropherograms of CFX spiked with its impurity A, NFX spiked with its impurity A and OFX spiked with its impurities A and E. These impurities were commercially available as pure compounds and thus, the peaks could be assigned. According to the certificate of analysis which was delivered together with “Ciprofloxacin for peak identification CRS” it was indicated that the HPLC method from the Ph. Eur. [24] is able to resolve four impurities (i.e. B, C, D and E) from the API. Of note, our proposed method was able to separate a total of eight impurities plus the main substance (Supporting Information Fig. S2A), thus offering a striking advantage in the sense that only one analysis run has to be performed to assess this drug. This is not the case in the methods which have been added to the Ph. Eur. and the International Pharmacopoeia (Supporting Information Figs. S1A and B) [8, 10]. In analogy, additional unknown signals could be observed in the reference standards of OFX impurity E, NFX impurity A and NFX itself (Supporting Information Fig. S3).
3.3.2 Linearity and accuracy Evaluation of linearity for CFX, NFX, MFX, OFX and LFX was carried out via plotting the corrected peak area ratios against an internal standard in the concentration range 0.04– 0.48 mg/ml. For all five FQs, a good linear relationship was
Figure 5. Overlaid electropherograms of CFX and NFX spiked with their respective impurity A as well as OFX spiked with its impurities A and E using the optimized MEEKC method.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.electrophoresis-journal.com
CE and CEC
Electrophoresis 2015, 36, 2736–2744
4 Concluding remarks
Table 2. Method accuracy summary for the five FQs at three levels (lower, nominal and upper)
MFX
CFX
NFX
LFX
OFX
Level (%)
Added (mg/ml)
Observed (mg/ml)
Accuracy (%)
sd
rsd (%)
C1–80 C2–100 C3–120 C1–80 C2–100 C3–120 C1–80 C2–100 C3–120 C1–80 C2–100 C3–120 C1–80 C2–100 C3–120
0.32 0.40 0.48 0.31 0.40 0.48 0.32 0.40 0.47 0.32 0.41 0.48 0.32 0.39 0.47
0.32 0.39 0.46 0.30 0.39 0.47 0.32 0.39 0.47 0.31 0.40 0.47 0.32 0.39 0.47
98.47 98.00 96.72 97.45 97.77 99.48 100.58 97.83 99.36 99.36 98.36 97.60 100.60 100.84 100.08
1.78 1.64 1.93 2.92 1.75 2.66 2.43 2.22 2.78 2.78 2.18 1.25 1.73 1.17 1.66
1.81 1.67 1.99 3.00 1.79 2.68 2.41 2.27 2.80 2.80 0.02 0.01 1.71 1.16 1.66
Table 3. Repeatability and intermediate precision
Repeatability (n = 6)
MFX CFX NFX LFX OFX
Intermediate precision (n = 12)
Mean corrected peak area/IS
sd
rsd (%)
Mean corrected peak area/IS
sd
rsd (%)
0.53 0.93 0.55 1.61 1.68
0.01 0.03 0.01 0.03 0.04
2.32 3.15 2.45 1.64 2.27
0.53 0.55 1.61 0.93 0.99
0.01 0.01 0.03 0.03 0.02
2.23 2.67 2.04 3.15 1.54
obtained. The regression data were y = 0.008x – 0.002 (R = 0.994) for MFX, y = 0.017x + 0.002 (R2 = 0.997) for CFX, y = 0.016x – 0.0001 (R2 = 0.997 for NFX), y = 0.009x + 0.020 (R2 = 0.989) for OFX and y = 0.009x + 0.020 (R2 = 0.989) for LFX, respectively. Alternatively, the ANOVA T- and F-test at a desired significance of 0.05 gave p-values below 0.001, for all five substances. H0 :1 is rejected, implying that a linear relation does exist between concentration and the responses. The percentage accuracies for the determination assay of CFX, NFX, OFX and LFX for the proposed generic MEEKC method are summarized in Table 2. 2
3.3.3 Precision The repeatability and intermediate precision data for the selected FQs are presented in Table 3. The values were calculated by using the mean corrected peak area ratio to the internal standard, expressed in terms of %rsd and were found to range from about 1 to 3%, suggesting the method has an adequate level of precision for the assay of the target drugs. C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
2743
A generic stability-indicating MEEKC method has been developed and validated. The protocol is able to separate five FQs in a mixture, or the respective impurities from the individual FQ within 20 min in a single run. The proposed MEEKC method is linear, simple, precise and accurate. It was also possible to separate impurities A, B, C, D and E of CFX and impurities A and E of OFX within one run which is not possible when using the official Ph. Eur. method. The method is representing an “all-in-one” concept which we anticipated before, meaning that for a testing laboratory the amount of required chemicals and supplies can be minimized [47]. This is a striking advantage in a resourceconstrained setting. The method was furthermore used to study the degradation behaviour of the five FQs in different media indicating that it is suitable for evaluating the stability of a drug sample. The authors gratefully thank the Alexander von HumboldtFoundation (Bonn, Germany) for awarding the Georg Forster Research Award for Senior Research Fellows to Dr. Eliangiringa Kaale. The Muhimbili University of Health and Allied Sciences is highly acknowledged for enabling the sabbatical leave of Dr. Eliangiringa Kaale. The authors declare no conflict of interest.
5 References [1] Anderson, R., Groundwater, P. W., Todd, A., Worsley, A. J., Antibacterial Agents: Chemistry, Mode of Action, Mechanisms of Resistance and Clinical Applications, Wiley, Chichester 2012. [2] Hooper, D. C., Biochim. Biophys. Acta 1998, 1400, 45–61. [3] Wolfson, J. S., Hooper, D. C., Clin. Microbiol. Rev. 1989, 2, 378–424. [4] Borcherding, S. M., Stevens, R., Nicholas, R. A., Corley, C. R., Self, T., J. Fam. Pract. 1996, 42, 69–78. [5] Huang, J., Kaul, G., Cai, C., Chatlapalli, R., HernandezAbad, P., Ghosh, K., Nagi, A., Int. J. Pharm. 2009, 382, 23–32. [6] Yu, L. X., Pharm. Res. 2008, 25, 781–791. [7] Yu, L. X., Amidon, G., Khan, M. A., Hoag, S. W., Polli, J., Raju, G. K., Woodcock, J., AAPS J. 2014, 16, 771–783. [8] European Pharmacopoeia, 8th Edition, Monograph “Ciprofloxacin hydrochloride” No. 04/2015:0888, European Directorate for the Quality of Medicines & HealthCare, Strasbourg, France 2014. [9] European Pharmacopoeia, 8th Edition, Monograph “Norfloxacin” No. 04/2011:1248, European Directorate for the Quality of Medicines & HealthCare, Strasbourg, France 2014. [10] The International Pharmacopoeia, 4th Edition, Health Organization, World Health Organization, Geneva, Switzerland 2010.
www.electrophoresis-journal.com
2744
E. Kaale et al.
[11] Miseljic, B., Popovic, G., Agbaba, D., Markovic, S., Simonovska, B., Vovk, I., J. AOAC Int. 2008, 91, 332– 338. [12] Cazedey, E. C. L., Salgado, H. R. N., Adv. Anal. Chem. 2012, 2, 74–79. [13] Natraj, K. S., Suvarna, Y., Prasanti, G., Saikumar, S. V., Int. Res. J. Pharm. 2013, 4, 178–181. [14] Argekar, A. P., Sawant, J. G., J. Planar Chromatogr. 1999, 12, 202–206. [15] Novakovic, J., Nesmerak, K., Nova, H., Filka, K., J. Pharm. Biomed. Anal. 2001, 25, 957–964. [16] Singh, B. K., Parwate, D. V., Srivastava, S., Shukla, S. K., Pharma. Chem. 2010, 2, 178–188. [17] Nyamweru, B. C., Kaale, E., Manyanga, V. P., Chambuso, M., Layloff, T., J. Planar Chromatogr. 2013, 26, 370– 374.
Electrophoresis 2015, 36, 2736–2744
[28] Li, L., Xia, Z., Yang, F., Chen, H., Zhang, Y., J. Sep. Sci. 2012, 35, 2101–2107. [29] Cruz, L. A., Hall, R., J. Pharm. Biomed. Anal. 2005, 38, 8–13. [30] Novakovic, J., Nesmerak, K., Nova, H., Filka, K., J. Pharm. Biomed. Anal. 2001, 25, 957–964. [31] Krzek, J., Hubicka, U., Szczepanczyk, J., J. AOAC Int. 2005, 88, 1530–1536. [32] Schulman, J. H., Stoeckenius, W., Prince, L. M., J. Phys. Chem. 1959, 63, 1677–1680. [33] McEvoy, E., Marsh, A., Altria, K., Donegan, S., Power, J., Electrophoresis 2007, 28, 193–207. [34] Ryan, R., Donegan, S., Power, J., McEvoy, E., Altria, K., Electrophoresis 2009, 30, 65–82. [35] Ryan, R., McEvoy, E., Donegan, S., Power, J., Altria, K., Electrophoresis 2011, 32, 184–201.
[18] Aksoy, B., Kucukguzel, I., Rollas, S., Chromatographia 2007, 66, S57-S63.
[36] Marsh, A., Clark, B., Broderick, M., Power, J., Donegan, S., Altria, K., Electrophoresis 2004, 25, 3970–3980.
[19] Lalitha Devi, M., Chandrasekhar, K. B., J. Pharm. Biomed. Anal. 2009, 50, 710–717.
[37] De Lu, J., Yuan, W., Fu, X. Y., Chin. Chem. Lett. 2001, 12, 155–156.
[20] Mehta, J., Pancholi, Y., Patel, V., Kshatri, N., Vyas, N., Int. J. PharmTech Res. 2010, 2, 1932–1942.
[38] Schwuger, M.-J., Stickdorn, K., Schomaecker, R., Chem. Rev. 1995, 95, 849–864.
[21] Reddy, B. V., Kumar, A. P., Reddy, G. V. R., Sahai, M., Sreeramulu, J., Park, J. H., Anal. Lett. 2010, 43, 2653–2662.
[39] Lamalle, C., Djang’Eing’A, M. R., Debrus, B., Lebrun, P., Crommen, J., Hubert, P., Servais, A.-C., Fillet, M., Electrophoresis 2012, 33, 1669–1678.
[22] Shervington, L. A., Abba, M., Hussain, B., Donnelly, J., J. Pharm. Biomed. Anal. 2005, 39, 769–775.
[40] The United States Pharmacopoeia USP 37 - NF 32, The United States Pharmacopoeial Convention, Rockville, MD, USA 2014.
[23] Faria, A. F., de Souza, M. V. N., de Oliveira, M. A. L., J. Braz. Chem. Soc. 2008, 19, 389–396. [24] European Pharmacopoeia, 8th Edition, Monograph “Ofloxacin” No. 01/2011:1455, European Directorate for the Quality of Medicines & HealthCare, Strasbourg, France 2014. [25] European Pharmacopoeia, 8th Edition, Monograph “Ciprofloxacin” No. 04/2011:1089, European Directorate for the Quality of Medicines & HealthCare, Strasbourg, France 2014.
[41] Guideline Q2(R1): Validation of Analytical Procedures: Text and Methodology, International Conference on Harmonisation, 2006, http://www.ich.org ´ S., Horvat, A. J. M., Mutavdˇzic´ Pavlovic, ´ D., [42] Babic, ˇ Kastelan-Macan, M., TrAC Trends Anal. Chem. 2007, 26, 1043–1061. [43] Baeuml, F., Welsch, T., J. Chromatogr. A 2002, 961, 35–44. [44] Chang, C. W., Chen, Y. C., Liu, C. Y., Electrophoresis 2014, 35, 2901–2906.
[26] The United States Pharmacopoeia USP 37 – NF 32, Monographs “Ciprofloxacin” and “Ciprofloxacin hydrochloride”, The United States Pharmacopoeial Convention, Rockville, MD, USA 2014.
[45] Pomponio, R., Gotti, R., Luppi, B., Cavrini, V., Electrophoresis 2003, 24, 1658–1667.
[27] Al Azzam, K. M., Saad, B., Adnan, R., Aboul-Enein, H. Y., Anal. Chim. Acta 2010, 674, 249–255.
[47] Hoellein, L., Holzgrabe, U., J. Pharm. Biomed. Anal. 2014, 98, 434–445.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[46] Marsh, A., Clark, B. J., Altria, K. D., J. Sep. Sci. 2005, 28, 2023–2032.
www.electrophoresis-journal.com
