2.3.1 Standard and test solutions for assay

The standard solution of RIV (0.5 mg mL⁻¹) was prepared in a solvent containing mixture of EtOH and 0.1% acetic acid (50:50, % v/v), by dissolving the reference substance in US bath for 30 min. The resolution solution (Res. sol.) containing 0.2 mg mL⁻¹ RIV, 0.4 μg mL⁻¹ impurity (imp.) G and 0.4 μg mL⁻¹ imp. H was prepared in EtOH/water mixture (40:60, % v/v). For the test solution preparation, twenty tablets containing 10 mg RIV were weighed and triturated in a clean dry mortar. Tablet mass corresponding to 10 mg RIV was transferred in 50 mL volumetric flask, dissolved in a mixture of EtOH and 0.1% acetic acid (50:50, % v/v) and treated on an ultrasonic bath for 15 min. 5.0 mL of this solution were diluted to a volume of 10.0 mL using EtOH/water mixture (40:60, % v/v). The placebo solutions for each medicinal product were prepared in the same manner as the test solution, by using placebo mixtures. Standard and test solutions for related substances quantification.

2.3.2 Standard and test solutions for related substances

Standard stock solution of RIV (0.2 mg mL⁻¹) was prepared in a solvent containing a mixture of EtOH and 0.1% acetic acid (50:50, % v/v) in the same manner as the preparation of standard solution of RIV for assay determination. Stock solutions of imp. G, imp. 14 and imp. H (0.2 mg mL⁻¹ each) were also prepared in solvent containing mixture of EtOH and 0.1% acetic acid (50:50, % v/v) and treated on an ultrasonic bath for 15 min. Impurity stock solutions were further diluted with EtOH/water mixture (40/60, % v/v) to obtain final concentration of 0.4 μg mL⁻¹, which is the working concentration of the method for each impurity (0.2% of working concentration, corresponding to the Ph. Eur. limits for unspecified impurities of RIV). The system suitability solution (Res. sol.) was the same as for the assay determination.

For the preparation of the test solutions, twenty tablets containing 10 mg RIV were weighed and triturated in a clean dry mortar. Tablet mass corresponding to 20 mg RIV was transferred in 50 mL volumetric flasks, solvent mixture of EtOH/0.1% acetic acid (50:50, % v/v) was added and the solution was treated on a magnetic stirrer for 15 min, and 30 min on an ultrasonic bath. The solution was filled up to volume with the same solvent. A part of this suspension was centrifuged for 15 min at 4,000 rpm. 5.0 mL of the supernatant were transferred in 10 mL volumetric flask and diluted up to volume with EtOH/water mixture (40/60, % v/v). The placebo solutions for each medicinal product were prepared in the same manner as the test solution, by using placebo mixtures.

2.5.1 DoE plan for related substances method optimization

During method optimization for related substances determination two sets of experiments were planed using DoE. The first set of experiments for screening the conditions was performed using Fractional Factorial DoE at two levels (11 experiments) with four experimental factors (results not shown): the percentage of EtOH in the isocratic part of the gradient (C₀ EtOH from 5 to 10% v/v), the length of the isocratic part of the gradient defined as gradient delayed time (t_D from 1 to 5 min), gradient time (t_G from 8 to 12 min) and final percentage of EtOH in the mobile phase (C₁ EtOH from 36 to 40%, v/v).

The second set of experiments (2³ Central Composite Face Design of Experiments, CCF DoE) was performed for the final optimization of the gradient elution, that required 17 experiments (2^k+2k+n experiments, where k was the number of parameters studied and n was the number of central points included, n = 3). Three repetitions at the central points are required to determine the experimental error variance and test the predictive validity of the model [20]. Chromatographic behavior was evaluated through three factors: C₀ EtOH (8–12% v/v), t_G (8–12 min) and C₁ EtOH (40–50% v/v), as shown in Table 1.

2.5.2 DoE plan for robustness testing for assay and related substances method

The robustness of both methods (for assay and related substances) was evaluated using 2³ Central Composite Face DoE, where small changes of values for the selected experimental factors, respectively critical for both methods, were made (Table 2).

Critical experimental factors for the assay method were: C₁ EtOH (35 ± 2%, v/v), flow rate (0.8–1.2 mL min⁻¹) and column temperature (45 ± 2 °C). The evaluated critical factor for related substances methods were: t_D (3 min ±1 min), t_G (8 min ±1 min) and flow rate (0.9–1.1 mL min⁻¹), as presented in Table 2.

3.1.1 Method development

Green analytical chemistry (GAC) principles which initiated a shift of attitudes and behavior in the chemical and pharmaceutical industry, can be seen as a critical tool for achieving sustainability [23]. As a multistep approach, it starts from the selection or modification of the appropriate analytical method that should meet the specified performance criteria, followed by the use of less solvents and/or less toxic solvents in the sample preparation and measurements. According to the physico-chemical properties, especially solubility tests, several solvents were evaluated in order to dissolve RIV [24]. RIV is practically insoluble in anhydrous EtOH, while it is freely soluble in DMSO [13]. However, during the chromatographic analysis of DMSO solutions, several peaks appeared that interfered with the peaks from RIV impurities. The RIV standard was found to be freely soluble in solvent containing mixture of EtOH and 0.1% acetic acid (50:50, % v/v). The selection of mixture of EtOH and acetic acid resembles the mobile phase composition and is much eco-friendlier than the solvents for sample preparation described in literature [13–18].

Considering the GAC principles, the starting point of the method development was the replacement of the acetonitrile in the mobile phase that is mostly used in the methods described in the literature with EtOH, as a more eco friendlier organic solvents [16–19, 23]. Similar retention time of RIV was observed with EtOH-based mobile phase (Rt 4.3 min), compared to ACN-based mobile phase (Rt 3.37 min) [23]. Although these chromatographic conditions satisfy the common system suitability requirements (retention time, symmetry factor, number of theoretical plates), still the method didn't have the required specificity regarding the imp. H as its closest eluting impurity. The resolution between the RIV and imp. H (Rs RIV/imp. H), using mobile phase of EtOH :H₂O (40:60%, v/v) was lower than 1.5. Therefore, the percentage of EtOH was decreased to 35% (v/v), which resulted in acceptable system suitability (Rs RIV/imp. H > 1.5), as well as improved greenness of the method. The Rs between imp. G and RIV (Rs imp. G/RIV) for assay determination is not critical, because the imp. G is well separated from the RIV peak (relative Rt imp. G is 0.6). Representative chromatograms for assay determination obtained under final conditions are presented in Fig. 1.

View Full Size

Chromatogram for assay determination under final conditions: a) Placebo solution b) Res. sol. c) Standard sol. and d) Test sol

Citation: Acta Chromatographica 36, 4; 10.1556/1326.2023.01178

• Download Figure
• Download figure as PowerPoint slide

3.1.2 Method validation

Validation of the analytical methods included the determination of method specificity, linearity, accuracy, precision, detection limit, and quantification limit according to ICH guideline [11].

Specificity/selectivity and system suitability: The system suitability was assessed by the following parameters: retention time of RIV (Rt RIV), peak symmetry (As factor 0.8–1.8), number of theoretical plates per column (N > 2000) and Rs RIV/imp. H ≥ 1.5. The obtained values for Rt RIV (5.67 min), N (7,928), As (1.2) and the resolution between the critical pair of peaks of 2.1, confirm the suitability of the proposed method. The specificity of the method was evaluated by injection of blank (solvent), placebo, resolution solution, standard solution of RIV, test solution, as well as separate standard solutions of its impurities (imp. G and imp. H). No interfering peaks were observed with the retention time of RIV: none of the examined impurities eluted at the retention time of RIV and there were no interfering peaks from the placebo (Fig. 1).

Linearity, accuracy and precision: Linearity of the method was evaluated from the five standard solutions prepared in the concentration range from 0.05 mg mL⁻¹ to 0.15 mg mL⁻¹ (50 – 150% of the working concentration). The obtained value of the correlation coefficient R² (0.9999) confirm the linearity of the method in the evaluated range. The RSD of the peak areas of the lowest and the highest concentration of the calibration curve obtained from six determinations for the studied range was 0.25% and 0.27%, respectively. The accuracy of the method was evaluated on spiked solutions of RIV at three concentration levels (70%, 100% and 130% of the working concentration). The confirmed accuracy of the method (Table 3) in the wider accuracy range (70%–130% instead of 80%–120%) confirms that the proposed method could be used for content uniformity determination as well. System precision and method precision were confirmed from six replicate injections of RIV standard at 100% working concentration (RSD = 0.34%) and six test solutions (RSD = 0.34%).

Robustness testing: The critical experimental responses (Rt RIV, Rs imp. G/RIV and Rs RIV/imp. H) from the robustness testing using the 2³ CCF DoE are presented in Table 2. The response surface plot from the robustness investigation is given in Fig. 2a. In all of the cases of the deliberately varied chromatographic conditions, the monitored system suitability parameters were within the acceptance criteria, confirming the robustness of the method.

View Full Size

Response contour plot of robustness testing for: a) assay method, b) related substances method

Citation: Acta Chromatographica 36, 4; 10.1556/1326.2023.01178

• Download Figure
• Download figure as PowerPoint slide

3.2.1 Method development

Green, ethanol-based RP-HPLC method for determination of related substances of RIV in tablets was developed. According to official methods for determination of related substances in rivaroxaban tablets described in Ph. Eur [13] and the USP pharmacopeia [14], impurity G and impurity H are crucial for demonstrating the system suitability. Considering the structural characteristics of imp. 14, which provide different polarity of this impurity compared to imp. H, in addition to specified impurities, imp. 14 was also included in this study. This approach provides more comprehensive evaluation of the chromatographic behavior of impurities with diverse polarities during the optimization of the method. Forced degradation studies were performed for the optimization of the method for determination related substances. The mixed solution was used to enable identification of all possible degradation products that could be produced in addition to imp. G, imp. H and imp. 14., thus improving the resolving capacity of the method. Taking into account the inherent stability of the molecule under influence of neutral hydrolysis, photolysis, thermolysis and oxidative forced degradation, these conditions were not considered for chromatographic optimization [16]. Four degradation products were obtained from force degradation under acidic conditions, with domination of imp. 14 and imp. H. Under alkaline conditions, a total of five degradation products were observed, including four unknown impurities (with one dominant early eluting impurity, assign as imp. 1) and the imp. H.

The initial mobile phase composition was selected based on comprehensive literature search and replacing most commonly used acetonitrile and water (or buffer) with EtOH and water respectively, as mobile phase eluents, aiming towards development of a green method. In order to obtain optimal separation between RIV and its related substances with a minimal number of experiments, the gradient elution conditions were evaluated using two sets of screening experiments: Fractional Factorial DoE (first set) and Central Composite Faced DoE (second set). The first DoE set was used to assess the effects of different chromatographic conditions (percentage of EtOH in the isocratic part of the gradient, C₀ EtOH; the length of the isocratic part of the gradient defined as gradient delayed time, t_D; gradient time, t_G; and final percentage of EtOH in the mobile phase, C₁ EtOH) on desired chromatographic responses: resolution between RIV and imp. H (Rs RIV/imp.H) and the capacity factor of the early eluting peak (k’ imp. 1). However, under investigated conditions the Rs RIV/imp.H was below 1.5 and the k’ imp. 1 was below 0.5. In all experiments, the resolution between imp. G and RIV was more than 5, as required for system suitability according to Ph. Eur, thus the response of this parameter was not considered as critical. In addition, there was no interference between the imp. 14 and other degradation peaks obtained with forced degradation. The evaluation of the critical responses obtained from the first DoE set, revealed that the delayed gradient time had no significant influence on the responses, therefore in the second DoE this experimental factor was excluded. The value of delayed gradient time was selected to be 3 min in the gradient program of the method, in order to eliminate the influence of the different dwell volumes in different HPLC instruments, enabling easy gradient method transfer between the different laboratories.

Central Composite Faced DoE was used for further optimization of the factors that have the highest effect on the chromatographic response (C₀ EtOH, t_G and C₁ EtOH). Seventeen experiments were performed to reveal the optimal chromatographic conditions for desired method performance: Rt RIV lower than 15 min, Rs RIV/imp. H higher than 1.5 and suitable retention of the early eluting peak (k’ imp. 1 more than 0.5). The assessment of the critical chromatographic responses (Rt, Rs RIV/imp. H and the k’ imp. 1) (Table 1) showed that satisfactory chromatographic performance related to the Rt and Rs could be easily achieved in a wide range of combinations of investigated experimental factors, while satisfying the criteria for k’ imp. 1 could be obtained in a very narrow area of experimental factors (Fig. 3). Rt RIV lower than 15 min and Rs RIV/imp. H higher than 1.5 could be achieved with lower percentage of EtOH in the mobile phase (C₁ ∼ 41%) and shorter gradient time (t_G ∼ 8 min) or with higher percentage of EtOH and longer gradient time. Suitable retention of early eluting peak (k’ imp. 1–1) is observed only if C₁ is between 40 and 41% and the t_G ranges between 8 and 9 min (Fig. 3). The C₀ EtOH did not influence critical chromatographic responses and it was set to 8%. After careful evaluation of the obtained results, the final gradient elution program was established.

View Full Size

Response contour plot obtained with the 2³ Central Composite Faced DoE for the related substances method optimization

Citation: Acta Chromatographica 36, 4; 10.1556/1326.2023.01178

• Download Figure
• Download figure as PowerPoint slide

3.2.2 Method validation

Specificity/selectivity and system suitability: The system suitability was evaluated by injection of the resolution solution. The method was found to be suitable for use because the Rs imp. G/RIV was 17.6 (≥5.0), the Rs RIV/imp.H was 2.1 (≥1.5) and the Rt for RIV was 14.9 min. The system suitability results imply that the green method developed in this study fulfills the system suitability criteria. Forced degradation studies were performed to assess the specificity/selectivity of the developed method for related substances determination. Representative chromatograms of diluent, placebo, each impurity solution at the specification limit (0.2%) and neutralized solutions obtained after acidic and alkaline hydrolysis are presented in Fig. 4. The peaks from all impurities are separated from each other and from RIV, without interfering peaks from the diluent or placebo, therefore the method was found specific and selective for the quantification of RIV related substances in tablets. In addition, peak purity analysis was conducted for all obtained peaks from chromatograms of the forced degradation solutions (Fig. 4a and 4b) in the wavelength range from 210 to 400 nm, using ChemStation for LC3D software. The peaks were considered to be spectrally pure if the peak purity factor value was above the set purity threshold of minimum 990. The absorbance threshold was set on 1 mAU for all relevant peaks and 0.5 mAU for the imp. 2 peak in the acidic degradation and imp. 3 peak in the alkaline condition. The obtained peak purity factor values for RIV (999.6), imp.H (999.8), imp. 14 (999.7) and for the unspecified impurities (the lower value was 993.7 for imp.1 in acidic condition) showed that the peaks are spectrally pure.

View Full Size

Representative chromatograms for the related substances method under final conditions: a) St. solution of RIV degraded under acidic condition, b) St. solution of RIV degraded under alkaline condition, c) Placebo, d) Resolution solution, e) St. sol. imp. 14, f) St. sol. imp. G, g) St. sol. imp. H, and h) Test solution

Citation: Acta Chromatographica 36, 4; 10.1556/1326.2023.01178

• Download Figure
• Download figure as PowerPoint slide

Linearity, accuracy and precision: Five solutions (prepared from the standard substances) at concentration levels ranging from quantification limit (QL) to 150% of the working concentration (0.05, 0.1, 0.15, 0.2, 0.4 and 0.6 μg mL⁻¹ for RIV and for each impurity) were used for evaluation of linearity of the method. The results from the linear regression analysis (Table 4) confirm the linearity of the method in the specified range. The obtained results for the analytical recovery in the entire linearity range (concentration levels: LOQ, 50%, 100% and 150% of the working concentration) of RIV, and the imp. G, imp. 14 and imp. H in spiked placebo (Table 3) confirmed the accuracy of the method. The repeatability of the system was confirmed by the obtained values for RSD (≤5%) of the peak area for six replicate injections of standard solutions containing RIV and impurities (imp. G, imp. H and imp. 14). RSD values for each detected and quantified related substance from analysis of 6 different test solutions spiked with the investigated impurities at a concentration level of 0.2% was below 5%, confirming the method repeatability.

Sensitivity: The detection and quantification limits (DL & QL) for RIV and RIV imp. G, imp. H and imp. 14 were calculated using the two different approaches: SD of the linear response and the slope of the curve, and the signal-to-noise approach (Table 4). The method shows satisfactory sensitivity for the intended use of the method, allowing quantification of the investigated impurities in the concentration range of 0.10 μg mL⁻¹ to 0.15 μg mL⁻¹.

Robustness testing: A rational assessment of the method robustness was achieved using the 2³ CCF DoE. The deliberate changes in the tested experimental factors didn't influence the critical resolution between RIV/imp. H (Table 2). The obtained response contour plots for the related substances method (Fig. 2b) show that deliberate alterations didn't affect method performance, confirming the robustness of the method.

3.3.3 Application of the proposed green RP-HPLC methods

The proposed ethanol-based HPLC methods were applied for assay and determination of related substances in rivaroxaban 10 mg tablets obtained from three different manufacturers. The content of RIV in tablets was calculated using the external standard method, whereas the percentage of the detected related substances was calculated using the peak area of RIV in diluted test solution (0.4 μg mL⁻¹) [13]. The obtained results (Table 5) confirm that the proposed methods are suitable for their intended purpose.

3.3.4 Greenness assessment of the proposed green RP-HPLC methods

The objective evaluation of greenness of the developed ethanol-based methods for determination of content and related substances of RIV in tablets was performed using the two quantitative tools ESA and AGREE. The penalty points for ESA score for both methods were calculated taking into account the amount and the toxicity of the solvents, the used technique, as well as the amount of waste generated from the whole procedure (sample preparation and instrumental analysis). The obtained ESA score for the assay method and for the related substances method were found to be 86 and 85, respectively (Table 6), showing excellent greenness of the proposed methods.

Table 6.

Analytical eco-scale and AGREE score for the proposed ethanol-based HPLC methods

The AGREE metric was applied to assess the compliance of each segment of the proposed methods with the twelve GAC principles. The results confirmed excellent agreement with the GAC p. 2, p. 4, p. 6, p. 10, p. 11 and p. 12 (green color in the pictograms) of both methods. Deviation from the GAC p. 1, p. 7 and p. 9 (expressed as orange colour in the pictograms) was observed due to the use of external sample pretreatment (p. 1); the amount of the analytical waste between 25 and 100 mL (p. 7); and the level of energy consumption, because LC is technique demanding high energy (p. 9). Considering that both methods are off-line, red color in the AGREE pictogram (non-compliance of the methods) is obtained for principle 3, as expected. The overall color of AGREE pictograms for both methods is pale green color (AGREE score of 0.66 for the assay method and 0.65 for related substances method) (Table 6). The results from assessment indicate that the proposed methods for determination of content and related substances of RIV in tablets, can be considered as eco-friendly.