Factorial design-assisted reversed phase-high performance liquid chromatography method for simultaneous determination of fluconazole, itraconazole and terbinafine

A 23 full factorial design model was used for the development of a new high performance liquid chromatography method with UV detection to estimate three antifungal drugs simultaneously. Fluconazole (FLU), itraconazole (ITR) and terbinafine (TRH) are co-administered for severe fungal infections. They have been determined using MOS-1 Hypersil C18 column and an isocratic eluent; methanol 95% and phosphate buffer 5% with 0.001% triethylamine. The pH was adjusted to 7, and the flow rate was 0.7 ml min−1. The three drugs were separated within less than 7 min at 210 nm. The developed method gave a linear response over 5–80 µg ml−1, 5–50 µg ml−1 and 1–50 µg ml−1 for FLU, ITR and TRH, respectively. It showed detection limits of 0.88, 0.29 and 0.20 µg ml−1 and quantification limits of 2.66, 0.88 and 0.60 µg ml−1 for the three drugs, respectively. The design of the experiment facilitated the optimization of different variables affecting the separation of the three drugs. The sensitivity of the designed method permitted the simultaneous estimation of ITR and TRH in spiked human plasma successfully.

In this study, a novel high performance liquid chromatography method is designed for the simultaneous analysis and quantitation of fluconazole, itraconazole and terbinafine, to our knowledge for the first time. The design of experiment (DOE) has been involved as an efficient methodology to test the influence of multiple factors which affect certain responses and the interaction between them with a low number of trials. Hence, it also participated in a decreasing amount of organic solvents and chemicals consumed during the study.  The purity percentage of ITR was 99%. -Organic solvents (HPLC grade) were obtained from Sigma-Aldrich (Germany).
Commercial dosage forms were obtained from pharmacies in Egyptian market including: -Flucoral ® capsules containing 150 mg fluconazole, product of Alfa Cure Pharmaceuticals Company, Cairo, Egypt; -Itrapex ® capsules containing 100 mg itraconazole, product of Global Napi Pharmaceuticals, Cairo, Egypt; -Lamisil ® tablets containing 250 mg terbinafine HCl, product of Novartis, Cairo, Egypt produced by Al-Andalus, Cairo, Egypt; and human plasma samples were provided by Mansoura University Hospitals and kept frozen at −20°C until used.

Standard solutions and mobile phase
Samples of each of FLU, ITR and TRH were accurately weighed and dissolved in methanol in 100 ml volumetric flasks to yield 100 µg ml −1 standard solutions. The mobile phase consists of methanol 95% and phosphate buffer 5% with 0.001%, v/v triethylamine adjusted at pH 7. The solution was subjected to ultrasonication and filtration through 0.45 µm membrane filters.

Constructing the calibration graphs
Aliquots of each of FLU, ITR and TRH standard solutions equivalent to final concentrations of 5.0-80.0, 5.0-50.0 and 1.0-50.0 µg ml −1 for the three drugs, respectively, were measured and transferred carefully into series of 10 ml volumetric flasks, completed with the mobile phase and mixed. Under optimum chromatographic sittings, 20 µl aliquots were injected in triplicate and the mobile phase flowed at a rate of 0.7 ml min −1 . The average peak area (y) was plotted against the final concentration of the drug (c) in µg ml −1 to construct the calibration curve for the analytes.

Analysis of pharmaceutical preparations
Separately, 10 capsules of each of Flucoral ® and Itrapex ® were evacuated and 10 tablets of Lamisil ® were ground, mixed and weighed precisely. A certain weight of each powdered drug was accurately transferred into a 100 ml volumetric flask to give concentrations equivalent to one capsule or tablet (250, 100 and 250 mg FLU, ITR and TRH, respectively), then about 40 ml of methanol was added. Each flask was subjected to sonication for 30 min, then the volumes were completed to 100 ml using methanol. The solutions were then filtered to get clear solutions. Working standard solutions were analysed as discussed in 'Constructing the calibration graphs'.

Analysis in spiking plasma samples
Using the proposed method, both ITR and TRH can be determined in human plasma. Aliquots of 1 ml of human plasma were transferred into a set of fixed capped tubes then increasing volumes of ITR and TRH standard solutions (final concentration reached: 5-13 µg ml −1 for ITR and 0.5-2 µg ml −1 for TRH) were added and mixed well. Methanol was used to complete the volume of each tube to 5 ml. Samples were subjected to vortex mixing for 15 s, then centrifugation at 3500 rpm for 30 min confirming the total separation of the drugs from plasma contents. The clear layer was filtered using syringe filters. Aliquots from the filtrate were accurately transferred into a 10 ml set of volumetric flasks and analysed by performing a blank experiment. The plasma content from the studied drugs is determined using regression equations.

. Choice of column
Three columns were investigated, which are: the Nucleosil 100-5 Phenyl column (250 mm × 4.6 mm i.d., 5 µm particle size), the Shimadzu VP-ODS C 18 column (250 mm × 4.6 mm i.d., 5 µm particle size) and the Thermo scientific ODS Hypersil C 18 column (250 mm × 4.6 mm i.d., 5 µm particle size) The latter was preferred as it gave distinct peaks with good resolution and short analysis time while the other columns gave longer analysis time.

Wavelength detection
Based on the UV spectra of the three drugs, 210 nm was selected as an optimum UV detection wavelength because they all show high absorbance at this wavelength, as shown in figure 2.

Mobile phase composition, pH and flow rate
Organic solvents as methanol, ethanol and acetonitrile were examined; it was found that the optimum one was methanol, as acetonitrile and ethanol induced overlapping of the two peaks of fluconazole and itraconazole. According to the prior trials, decreasing the ratio of methanol resulted in a major increase in the retention time of TRH, so 85-95% of methanol was investigated for the study.
The optimization of chromatographic conditions 'univariate optimization' is a complex process as it requires many experiments to achieve the optimum conditions. The study includes one variable at a time, while others remain constant which is time-consuming.
A full factorial design is a type of DOE 'multivariate optimization' which allows us to investigate the effect of all the factors with a simultaneous variation of them and to watch the responses encountered from independent factors and the interactions between those factors [37]. In this study, for the optimization of the chromatographic condition, 2 3 full factorial design was applied as it included two levels and three independent factors. The most important factors affecting the HPLC method were per cent of methanol, royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 8: 202130 flow rate and buffer pH and their optimization was carried out by DOE. Two different ratios of methanol such as 85 and 95 were studied; the pH of sodium hydrogen phosphate selected to be studied were 5 and 7. These selected pHs were reported in the analysis of acidic and basic drugs in a reversed phase (RP) system and the flow rates of 0.7 and 1 ml min −1 were chosen. Consequently, the two levels were (−1) for the lower level and (+1) for the higher level and the three independent factors were per cent of methanol (A), pH of buffer (B) and flow rate (C) [38].
The 2 3 full factorial design suggested eight experiments to analyse the interaction of each level on the responses, which were the resolution between fluconazole and itraconazole (R1), resolution between itraconazole and terbinafine (R2), tailing of terbinafine peak (R3) and number of the theoretical plate of TRH USP (NTP) (R4). The two levels, independent variables and dependent variables, are illustrated in tables 1 and 2.
The significance of independent factors was evaluated by means of the estimated Fisher statistical test for variance analysis (ANOVA) model [39], which is applied on the responses to study the effect of these independent factors on the responses and the interactions between them. The polynomial equation for the experimental design with three factors is given below: where R is the response, β is the regression coefficients and A, B and C represent per cent of methanol, pH of buffer and flow rate, respectively.
To ensure that the optimum conditions are obtained, MINITAB response optimizer calculates the composite desirability (D) which evaluates if the responses are in their acceptable limits and it ranges from zero to one. Zero is not accepted as it means that many of the responses are out of their accepted limits, while one means that the condition reached is optimum, so its value is better to be one or near one (table 2).  royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 8: 202130 A full factorial design resulted in the optimum solution; hence, the optimization plot (figure 3) shows how the composite desirability and responses are affected by the three factors and the interaction between them to reach the optimum condition.
The factorial design can also provide interaction plots (figure 4), which show that increasing methanol and pH and decreasing flow rate lead to increasing NTP and Rs1 and minimizing tailing and Rs2 which is confirmed by the main effect plots (figure 5). Pareto charts (figure 6) show that methanol% in the mobile phase (B) has a major effect on the chromatographic performance (Rs1 and Rs2) and has a statistically significant effect for a 95% confidence level, while pH (A) has the major effect on NTP. However, it is not potentially effective for a 95% confidence level. pH and methanol% (AB) have the strongest effect on tailing response.  royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 8: 202130

Validation parameters
Validation was carried out as guided by the International Conference on Harmonization (ICH) Q2R1 [41]. The linearity and ranges for the three drugs were investigated by the regression equations after analysis of six concentrations for each of the three drugs. The suggested HPLC method was applied over the ranges of (5-80, 1-50 and 1-50 µg ml −1 ), respectively, to pure samples of FLU, ITR and TRH, as shown in     royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 8: 202130 The limit of detection (LOD) and limit of quantitation (LOQ) were calculated according to the following equations [41], and the resulted data are shown in table 4: By applying statistical analysis [42], no significant difference was found after comparing the results obtained from the proposed method with those from the comparison methods [6,9,10], as represented in table 5. The comparison methods were spectrophotometric methods that depend on measuring the Table 5. Assay results for the determination of the studied drugs in pure form by the proposed and comparison methods. (Values between parentheses are the tabulated t and F values, respectively, at p = 0.05 [42]  The robustness of the proposed HPLC methods has been proved by making a slight variation in the chromatographic conditions such as pH (7 ± 0.2), methanol (95 ± 2%) and flow rate (0.7 ± 0.1 min ml −1 ). Such minor changes were found to have minimal effect on drug resolution, indicating good robustness of the method.

Application in tablets dosage forms and synthetic mixtures
The results obtained from the suggested method for evaluating FLU, ITR and TRH in commercial tablets and synthetic mixtures were compared with those obtained using previous methods [6,9,10]. By applying statistical analysis using Student's t-test and variance ratio F-test [42], no important difference regarding the accuracy and precision was found, as shown in tables 7 and 8.

Application in biological fluid
The suggested approach applied to determine ITR and TRH in spiked human plasma. The results obtained from spiked plasma are presented in table 9. Using the optimized experimental conditions, a linear relationship was constructed by plotting the peak area against the drug concentration.
Therapeutic concentration of ITR is greater than 0.25 mg l −1 and peak plasma levels of approximately 1 mg l −1 TRH occur 2 h after a single oral dose of 250 mg [43] and other reports indicate higher plasma concentration of ITR greater than 1.0 mg ml −1 in resistant oropharyngeal candidiasis [44].
Linear regression analysis of the data gave the following equation:   Fluconazole could not be determined in spiked human plasma by this method because of overlapping of its peak with the plasma peak, as shown in figure 8.

Conclusion
A factorial design-assisted HPLC method has been developed to separate three antifungal drugs, namely FLU, ITR and TRH. The proposed method is a straightforward one that saves time and money for reaching the optimum chromatographic conditions. The developed methodology permitted the whole separation to proceed in less than 8 min. Moreover, both ITR and TRH can be determined in spiked human plasma with satisfactory results. The proposed methodology is fast, simple and reproducible with a wide linear range in comparison with other previous reports for the analysis of the three drugs. It could be used for routine analysis of these antifungal drugs.
Ethics. Collection of biological samples has been reviewed and reapproved by the ethical committee of Faculty of Pharmacy, Mansoura University (code no. 2020-99).
Data accessibility. Data are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.sqv9s4n2t [45]. Authors' contributions. Laboratory work was carried out by A.R. in addition to data analysis and drafting the manuscript.
H.E. and S.S. contributed to the study design and the statistics. A.E.-B. organized the overall study. All authors permitted the publication of the manuscript. royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 8: 202130