Using an innovative quality-by-design approach for the development of a stability-indicating UPLC/Q-TOF-ESI-MS/MS method for stressed degradation products of imatinib mesylate

Abstract

A stability-indicating UPLC/Q-TOF-ESI-MS/MS method has been developed for the simultaneous determination of imatinib mesylate (IMM) and its impurity and degradation products in the active pharmaceutical ingredient (API) and drug products. Gradient elution of 0.1% g mL⁻¹ ammonium formate buffer at pH 8.57 and acetonitrile were used as the mobile phase and a Waters Acquity UPLC CSH C18, 100 mm × 2.1 mm, 1.7 μm particle size column was utilized as the stationary phase. Forced degradation, such as acid and base hydrolysis, and oxidative stress conditions of IMM were carried out to prove the stability-indicating performance of the method. The quality-by-design (QbD) principle was applied to the method development approach and the chromatography modeling software DryLab®2000 plus and Design Expert®8.0.6 were used to optimize the chromatographic method. The robustness study (Design Space) was performed by varying three critical chromatographic parameters (flow rate, temperature, and pH) at 3 levels (+1, 0, and −1). The result showed that baseline separation of all peaks of IMM and its impurity and degradation products could be achieved and a resolution of 2.0 could be reached in all experiments. The UPLC method was validated for specificity, linearity, accuracy, precision and robustness in compliance with the ICH guideline Q2 (R1). The method we developed was a fast, robust and reliable UPLC method with higher suitability and specificity. Furthermore, elemental composition and major fragments of the impurity and degradation products of IMM were characterized through the optimized UPLC/Q-TOF-MS/MS analysis.

---

1. Introduction

Imatinib mesylate (IMM) is a protein tyrosine kinase inhibitor. As an anticancer drug, it was often used to treat patients with Philadelphia positive chronic myeloid leukemia, chronic myeloid leukemia (CML) in blast crisis, gastrointestinal stromal tumors (GISTs), acute lymphoblastic leukemia and aggressive systemic mastocytosis. The drug was developed by the Swedish company Novartis.¹

Imatinib mesylate was designated chemically as 4-[(4-methyl-1-piperazinyl) methyl]-N-[4-methyl-3-[[4-(3-pyridinyl)-2-pyrimidinyl]amino]-phenyl]benzamide methanesulfonate (Fig. 1).

Fig. 1 Structure of imatinib mesylate (IMM).

At present, there is no international pharmacopoeia standard for the estimation of related substances and determination of IMM in the active pharmaceutical ingredient (API) and pharmaceutical formulations. Only several HPLC and UPLC procedures for the determination of IMM and its related substances have been reported in the literature.^1,2 Vivekanand et al.³ showed a validated HPLC method for IMM. Moreover, some RP-HPLC methods were used for estimation of IMM and its impurities.^4,5 Although HPLC is widely used to analyze the impurities in the API and pharmaceutical formulations, it also has some disadvantages, such as longer analysis time and larger consumption of organic solvents. UPLC as a new technique shows better separation capabilities, shorter separation time and less solvent consumption.¹

Nowadays, establishment of a validated stability-indicating assay method for the determination of related substances and IMM in the API and drug products is recommended by the International Conference on Harmonization (ICH) guidelines⁶ and the United State Pharmacopoeia (USP).⁷ Of these above methods, only one¹ is stressed to be a stability-indicating analytical method for IMM and its impurities and degradation products. As an important part of the process of drug product development, a stability test contributes to identifying the degradation products and impurities for the development of a stability-indicating method. Many environmental conditions significantly influence drug stability, such as heat, light, hydrolysis and oxidation. The degradation products of IMM under hydrolytic, oxidation conditions have been reported in the literature.^1,2 In this paper, the impurity and degradation products, namely Imp-A, B, C₁, C₂, D and E, are shown in Fig. 2.^8,9

Fig. 2 Chemical structures of impurity and degradation products of IMM. (Imp-A: 4-(4-methyl-piperazin-1-ylmethyl)-benzoic acid; Imp-B: 4-(4-methyl-1,4-dioxy-piperazin-1-ylmethyl)-N-[4-methyl-3-(4-pyridin-3-yl-pyrimidin-2-ylamino)-phenyl]-benzamide; Imp-C₁: 4-(4-methyl-1-oxy-piperazin-1-ylmethyl)-N-[4-methyl-3-(4-pyridin-3-yl-pyrimidin-2-ylamino)-phenyl]-benzamide; Imp-C₂: 4-(4-methyl-4-oxy-piperazin-1-ylmethyl)-N-[4-methyl-3-(4-pyridin-3-yl-pyrimidin-2-ylamino)-phenyl]-benzamide; Imp-D: N-(5-amino-2-methylphenl)-4-(3-pyridyl)-2-pyrimidine amine; Imp-E: 4-[(4-methylpiperazin-1-yl)methyl]-N-[4-methyl-3-[4-(3-pyridinyl)-2-pyrimidinyl]chloromethylamino-phenyl]-benzamide).

Hence, it is necessary to develop a simple, fast and reliable stability-indicating UPLC method for the determination of IMM and its impurity and degradation products in the API and pharmaceutical formulations.

In the past years, a trial-and-error approach was often used for the development of a chromatographic method, such as the one-factor-at-a-time (OFAT) method. The principle, which requires a great number of experiments to develop a relatively proper method, has many shortcomings, such as being time-consuming, solvent-wasting and having some unknown factors.¹⁰ Nowadays, experimental design principles have been widely and successfully used for method development in the pharmaceutical industry. In our study, a small number of experiments were needed to achieve an ideal result based on the chromatography modeling software DryLab®2000 plus which was used for UPLC method development. At the same time, the quality-by-design (QbD) principle, which is recommended by the ICH guideline Q8 (R2)¹¹ and widely requested in pharmaceutical development,¹² was applied to build a more scientific and risk-based multifactorial approach. The software Design Expert®8.0.6 was used in the robustness study which is a fundamental criteria of quality in a HPLC method.¹³ The optimum regions in which it had the best separation of each peak could be found out by the application of the software. Hence, a new stability-indicating UPLC/Q-TOF-ESI-MS/MS method was developed for the determination of IMM and its impurity and degradation products in the API and drug products.¹⁴ Moreover, elemental composition and major fragments of the impurity and degradation products of IMM were characterized through the optimized UPLC/Q-TOF-MS/MS analysis.^15,16

2. Experimental

2.1. UPLC equipment and chromatographic conditions

LC-MS/MS analysis was carried out on an Agilent 1290 series UPLC instrument (Agilent Technologies, USA) coupled to a quadrupole time-of-flight mass spectrometer (Bruker micrOTOF-QII, Germany) equipped with an electrospray ionization source. Two different columns, a Waters Acquity UPLC CSH C18 column (100 × 2.10 mm i.d., particle size 1.7 μm) and a Phenomenex kinetex C18 column (100 × 2.10 mm i.d., particle size 1.7 μm), were used in the study.

All chromatographic experiments were performed in a gradient mode. Solvent A was 0.1% g mL⁻¹ ammonium formate buffer at pH 8.57 (adjusted with ammonia solution) and solvent B was acetonitrile. The flow rate was set to 0.45 mL min⁻¹ and the injection volume was 2 μL. The temperature in the screening experiments was at 41.88 °C. UV detection was carried out at 269 nm. The gradient analysis was as follows: (T/%B) 0/5; 1/15; 1.1/25; 5/25; 5.1/40; 7/48; 9/48.

2.2. Drug and reagents

The API of IMM and impurities (standard substances, purity 99.00%) was provided by Tianjin Taipu pharmaceutical technology development Co., Ltd. Acetonitrile and methanol were HPLC grade, and were purchased from Tianjin KangKeDe Chemical Reagent Co., Ltd. Ammonium formate, ammonia solution, sodium hydroxide, hydrochloric acid and hydrogen peroxide were analytical grade, and all solvents were purchased from Tianjin Jiangtian Chemical Reagent Co., Ltd. High purity water was purified by a Millipore Milli-Q academic water purification system (Tianjin On-well Scientific Co., Ltd).

2.3. Software

The chromatography modeling software DryLab®2000 plus (Molnar-Institute, Berlin, Germany) was used for the optimization of gradient time, temperature and pH to separate a mixture of IMM and its impurity and degradation products. The robustness evaluation was performed by Design Expert®8.0.6 (Stat-Ease, Inc., Minneapolis, MN).

2.4. Preparation of drug product solution

A standard stock solution containing IMM and all six impurities was prepared with a mixer of MeOH and CH₃CN (60 : 40, v/v). The sample solution was prepared by dissolving the API of 5 mg IMM substance in a 10 mL volumetric flask with a mixer of MeOH and CH₃CN (60 : 40, v/v), diluted to obtain a concentration of 0.5 mg mL⁻¹. The resulting solution was filtered through a 0.22 μm nylon filter. This clear solution was used for the UPLC determination.

2.5. Stress studies

In the stress studies, the 0.5 mg mL⁻¹ concentration of IMM drug substance was used to provide an indication of the stability-indicating property and specificity of the proposed method. Degradation was attempted under stress conditions of acid (1.0 M HCl at 60 °C, 6 h), base (1.0 M NaOH at 60 °C, 6 h), and oxidation (1% H₂O₂ at 60 °C, 1 h) to evaluate the ability of the proposed method to separate IMM from its impurity and degradation products. All of the stress conditions were done in the dark. An UV detector (UPLC) and Q-TOF-ESI-MS/MS were used to check the peak purity of the stressed sample.

3. Results and discussion

3.1. Design of experiments

3.1.1. Development of an UPLC method for the simultaneous determination of IMM and its impurity and degradation products. Firstly, the mixed sample solution containing IMM and its impurity and degradation products was analyzed on a Waters Acquity UPLC CSH C18 column (100 × 2.10 mm i.d., particle size 1.7 μm), using 0.1% g mL⁻¹ ammonium formate buffer and acetonitrile in a gradient mode (the initial ratio was 95 : 5) at a flow rate of 0.5 mL min⁻¹ and the temperature was 35 °C. The results showed that the separation and peak shape of each peak in the mixed sample were not satisfactory.

So, a series of trials were done by changing the pH of the buffer solution (pH values of 3.0, 6.0 and 9.0), the flow rates (0.4 mL min⁻¹, 0.5 mL min⁻¹, and 0.6 mL min⁻¹), the detection wave length (237 nm, 254 nm, and 269 nm) and the stationary phase [Waters Acquity UPLC CSH C18 column (100 × 2.10 mm i.d., particle size 1.7 μm), Phenomenex kinetex C18 column (100 × 2.10 mm i.d., particle size 1.7 μm)]. After those kinds of trials, the preliminary UPLC method was obtained: on the Waters Acquity UPLC CSH C18 column at 35 °C using 0.1% g mL⁻¹ ammonium formate buffer and acetonitrile in a gradient mode (the initial ratio was 95 : 5) at a flow rate of 0.5 mL min⁻¹ and the UV detection was carried out at 269 nm. The choice of the buffer solution, which could also be used for the LC-MS study, expanded its applicability.

3.1.2. Development and optimization of Q-TOF-ESI-MS/MS parameters. The Q-TOF-ESI source conditions were also optimized, such as the drying gas temperature, drying gas flow, spray voltage and capillary voltage and so on to maximize the ionization of the source and the sensitivity. The impurity and degradation products could be identified and characterized even at a very low concentration under these conditions.¹⁷ After a series of attempts, the conditions were as follows:

In positive ion mode; drying gas temperature: 200 °C; drying gas flow: 6.0 L min⁻¹; nebulizer: 1.8 bar; capillary voltage: 4500 V; and scanning range: 50–1000 m/z.

Voltage parameters: transfer: funnel 1 RF: 200 Vpp; funnel 2 RF: 300 Vpp; hexapole RF: 300 Vpp; quadrupole: ion energy: 6 eV; low mass: 200 m/z; collision cell: collision energy: 10 eV; collision RF: 300 Vpp; transfer time: 50 μs; pre-pulse storage: 10 μs.

3.2. UPLC method optimization

3.2.1. Selection of working point. Three chromatographic parameters (gradient time (t_G), temperature (T) and pH) were the highly influential experimental parameters that had a significant influence on the selectivity in our study. So they were chosen as the working point instead of the less affecting parameters, such as the flow rate, ternary eluent composition and so on.

3.2.2. Optimization phase: optimization of measured parameters t_G, T, and pH. Initial input data were acquired under the following conditions: gradient times of 8 and 10 min, temperatures of 35 °C and 50 °C, and eluent A was 0.1% g mL⁻¹ ammonium formate buffer with pH values of 3.0, 6.0 and 9.0. According to the design of the experiments by DryLab®2000 plus, a number of experiments were produced, and the data were transferred to DryLab®2000 plus. In this study, the center point (t_G: 9 min, T: 42 °C, pH: 8.0) was found from the t_G–T model corresponding to three pH values and the t_G–pH model (Fig. 3). The results of the major parameters in the chromatogram are shown in Table 1. In this situation, several of the critical peak pairs (Rs, crit > 2.0) and each component had a good separation and peak shape.

Fig. 3 (A) Gradient time–temperature model and (B) gradient time–pH model for the same sample as described.

Table 1 The results tables of the experimental conditions of the measured parameters necessary for the generation of 3D models. (A) was the t_G–T model, and the pH was 9.0, and (B) was the t_G–pH model

(A)
#	Name	tR (min)	Area	Tail	Rs
1	Imp-A	1.00	15.1	1.00	44.09
2	Imp-B	3.08	1680.2	1.00	7.89
3	Imp-C1	3.7	721.1	1.00	7.85
4	Imp-D	4.44	24.8	1.00	3.67
5	Imp-C2	4.83	416.2	1.00	13.62
6	IMM	6.59	3123.4	1.00	2.88
7	Imp-E	7.02	31.4	1.00

---

(B)
#	Name	tR (min)	Area	Tail	Rs
1	Imp-A	0.87	33.3	1.00	61.97
2	Imp-B	3.72	6197.3	1.00	10.89
3	Imp-C1	4.61	1324.0	1.00	19.96
4	Imp-C2	6.72	1502.7	1.00	14.43
5	Imp-D	7.49	370.0	1.00	3.27
6	IMM	8.69	17 246.7	1.00

---

3.3. Multifactorial robustness study

3.3.1. Box–Behnken design. In our study, the Box–Behnken design was used to obtain a proper model for optimization of the largest possible robustness with three main variation values (flow rate, temperature and pH) at three levels.¹⁸ Within cube A, the three measured parameters’ tolerances were determined to be: flow rate (v): 0.5 ± 0.05 mL min⁻¹; temperature (T): 42 ± 5 °C; and pH: 8.0 ± 1.0 (Table 2). 17 new experiments were carried out according to the experimental design to explore the largest possible tolerances and the optimum conditions of the three main variation values. The predicted values were obtained by the model fitting technique in Design Expert®8.0.6 software after input of the response values, which were Rs₁ (Imp-B and Imp-C₁), Rs₂ (Imp-D and Imp-C₂), Rs₃ (IMM and Imp-E) and T_IMM.

Table 2 Three levels of factors used for the multifactorial robustness study

Factor name	Three levels
	Low (−1)	Middle (0)	High (+1)
T (°C)	37	42	47
v (mL min^−1)	0.45	0.50	0.55
pH	7.0	8.0	9.0

---

3.3.2. BBD modeling. Based on the Box–Behnken experimental design model, the polynomial equations generated in coded values are presented as follows:

Rs₁ (Imp-B and Imp-C₁) = 2.99 − 0.057A + 0.18C + 0.27AC + 0.21A² + 0.59C² (1)

Rs₂ (Imp-D and Imp-C₂) = 1.71 − 0.084A + 0.060B − 0.13C − 0.46AC − 0.36BC (2)

Rs₃ (IMM and Imp-E) = 2.46 − 0.29A + 0.071B + 1.17C + 0.30BC + 0.20A² + 1.01C² (3)

T_IMM = 0.97 + 0.053A − 0.011B − 0.089C − 0.14A² + 0.13B² + 0.98C² (4)

Pareto analysis of variance analysis (ANOVA) and regression analysis were used to test the fitness of the models. The results in Table 6 point out that the equation sufficiently presented the relationship between the input parameters and the response variables. ANOVA is a statistical technique that subdivides the total variation in a data set into component parts combined with sources of variation on the variables of the model.¹⁹ The ANOVA results in Table 3 show the F-value for Rs₁ (Imp-B and Imp-C₁), Rs₂ (Imp-D and Imp-C₂), Rs₃ (IMM and Imp-E) and T_IMM as 8.08, 1.82, 60.07 and 238.76 respectively, implying that the model was highly significant. The p-values that were lower than 0.05 indicated that the quadratic model was statistically significant.²⁰

Table 3 Pareto analysis of variance analysis (ANOVA) and statistical parameters of the model

Source	Coefficient estimate	Sum of squares	DF	Standard error	Mean square	F-value	p-value
Rs1 (Imp-B and Imp-C1)
Model		2.29	5		0.46	8.08	0.0020
A-v	−0.057	0.026	1	0.084	0.026	0.47	0.5089
C-pH	0.18	0.27	1	0.084	0.27	4.70	0.0531
AC	0.27	0.28	1	0.12	0.28	4.95	0.0480
A^2	0.21	0.19	1	0.12	0.19	3.28	0.0974
C^2	0.59	1.47	1	0.12	1.47	25.90	0.0003
Residual		0.62	11		0.057
Lack of fit		0.57	7		0.082	6.63	0.0433
Mean	3.37
C.V.%	7.08
Adeq precision	8.409
Rs2 (Imp-D and Imp-C2)
Model		1.58	5		0.32	1.82	0.1888
A-v	−0.084	0.056	1	0.15	0.056	0.32	0.5808
B-T	0.060	0.029	1	0.15	0.029	0.17	0.6914
C-pH	−0.13	0.14	1	0.15	0.14	0.79	0.3917
AC	−0.46	0.85	1	0.21	0.85	4.88	0.0493
BC	−0.36	0.51	1	0.21	0.51	2.95	0.1139
Residual		1.91	11		0.17
Lack of fit		1.80	7		0.26	9.89	0.0214
Mean	1.71
C.V.%	4.40
Adeq precision	4.947
Rs3 (IMM and Imp-E)
Model		16.63	6		2.77	60.07	<0.0001
A-v	−0.29	0.69	1	0.076	0.69	14.96	0.0031
B-T	0.071	0.041	1	0.076	0.041	0.88	0.3703
C-pH	1.17	10.95	1	0.076	10.95	237.34	<0.0001
BC	0.30	0.35	1	0.11	0.35	7.54	0.0206
A^2	0.20	0.17	1	0.10	0.17	3.79	0.0803
C^2	1.01	4.31	1	0.10	4.31	93.52	<0.0001
Residual		0.46	10		0.046
Lack of fit		0.45	6		0.075	26.60	0.0035
Mean	3.03
C.V.%	7.09
Adeq precision	22.207
TIMM
Model		4.28	6		0.71	238.76	<0.0001
A-v	0.053	0.022	1	0.019	0.022	7.38	0.0217
B-T	−0.011	1.012 × 10^−3	1	0.019	1.012 × 10^−3	0.34	0.5735
C-pH	−0.089	0.063	1	0.019	0.063	21.08	0.0010
A^2	−0.14	0.082	1	0.027	0.082	27.41	0.0004
B^2	0.13	0.074	1	0.027	0.074	24.91	0.0005
C^2	0.98	4.03	1	0.027	4.03	1347.15	<0.0001
Residual			10		2.990 × 10^−3
Lack of fit			6		4.963 × 10^−3	165.42	<0.0001
Mean	1.43
C.V.%	3.82
Adeq precision	36.518

---

The fitness of the models was also evaluated by calculation of the coefficient of determination (R²) and the adjusted R². The values of R² were calculated to be 0.7961, 0.5146, 0.9866 and 0.9941 for Rs₁ (Imp-B and Imp-C₁), Rs₂ (Imp-D and Imp-C₂), Rs₃ (IMM and Imp-E) and T_IMM respectively, which implied that most of the experimental results were well suited. The high value of the adjusted R² (0.5339 for Rs₁ (Imp-B and Imp-C₁), −0.1095 for Rs₂ (Imp-D and Imp-C₂), 0.9693 for Rs₃ (IMM and Imp-E) and 0.9865 for T_IMM) supported a high accuracy between the measured and the predicted values. The high coefficient of determination and very small p-value (<0.0001) demonstrated that the quadratic polynomial model was significant and adequate to characterize the actual relationship between the input and response. The coefficient of variation (CV) indicates the scattering of the experimental points from the predicted values of the second order polynomial model.²¹ The low coefficient of variation value (7.08, 4.40, 7.09 and 3.82) showed a high level of precision and reliability of the experiments conducted.

3.3.3. Effect of independent variables on the separation. Rs₁ (Imp-B and Imp-C₁), Rs₂ (Imp-D and Imp-C₂), Rs₃ (IMM and Imp-E) and T_IMM were set as the response parameters to obtain the best conditions of separation. The best parameters were carried out with various flow rates, temperatures and pH values to find out the optimum conditions. 3D response surface plots showed the influence of two variables on the process, whilst at the same time the third variable was kept constant. The surface plots representing the specific surface area of the response parameters versus independent variables are illustrated in Fig. 4–7.

Fig. 4 Response surface plots (3D) showing the effects of (a) flow rate and temperature, (b) temperature and pH and (c) flow rate and pH on Rs₁ (Imp-B and Imp-C₁).

Fig. 4 shows the effects of flow rate, temperature and pH on Rs₁ (Imp-B and Imp-C₁). Fig. 4(a) depicts that the effect of flow rate and temperature on Rs₁ (Imp-B and Imp-C₁). Fig. 4(b) illustrates the interaction effect of the independent variables (temperature and pH) on Rs₁ (Imp-B and Imp-C₁). The response surface plot (Fig. 4(c)) shows that the flow rate and pH also have an effect on Rs₁ (Imp-B and Imp-C₁). Each parameter showed a significant influence on Rs₁ (Imp-B and Imp-C₁). In each figure, we found that the value of Rs₁ (Imp-B and Imp-C₁) increased after a certain degree of decline at the beginning with the increase of flow rate, temperature or pH. However, the best value without considering the other three response parameters couldn’t be found. The same state also happened in the interaction effect of the independent variables (flow rate, temperature and pH) on Rs₂ (Imp-D and Imp-C₂) (Fig. 5). Fig. 5(a) shows the effect of flow rate and temperature on Rs₂ (Imp-D and Imp-C₂). The best value of Rs₂ (Imp-D and Imp-C₂) was located in the edge of the 3D response surface plots (Fig. 5(b) and (c)). Fig. 6 shows the influences of flow rate, temperature and pH on Rs₃ (IMM and Imp-E). Fig. 6(a) depicts that Rs₃ (IMM and Imp-E) has the most fitness with a low flow rate and high temperature. The response surface plots (Fig. 6(b) and (c)) show the interaction effect of the independent variables on Rs₃ (IMM and Imp-E). The variables exhibit a strong influence on Rs₃ (IMM and Imp-E). Fig. 7 depicts the effects of flow rate, temperature and pH on T_IMM. Fig. 7(a) shows that the flow rate and temperature have a complex effect on T_IMM. The response surface plots (Fig. 7(b) and (c)) illustrate that T_IMM has the most suitable figure with the interaction effect of the independent variables. In this study, the independent parameters showed a significant influence on the separation efficiency. The maximum value of each response parameter was usually achieved at a relatively high temperature and pH due to the structures of components and the theory of analysis. However, the optimized independent variables in the final method should be obtained by the desired function in the Design Expert®8.0.6 software.

Fig. 5 Response surface plots (3D) showing the effects of (a) flow rate and temperature, (b) temperature and pH and (c) flow rate and pH on Rs₂ (Imp-D and Imp-C₂).

Fig. 6 Response surface plots (3D) showing the effects of (a) flow rate and temperature, (b) temperature and pH and (c) flow rate and pH on Rs₃ (IMM and Imp-E).

Fig. 7 Response surface plots (3D) showing the effects of (a) flow rate and temperature, (b) temperature and pH and (c) flow rate and pH on T_IMM.

3.3.4. Selection of optimum conditions. Optimum conditions for the analytical method were produced to get the best separation of each component. The analytical method was optimized from varying the following parameters: (1) flow rate (0.45–0.55 mL min⁻¹), (2) temperature (37–47 °C) and (3) pH (7.0–9.0) respectively. These were set for the most desirable values. The methodology of the desired function was applied to obtain the optimized independent variables. It indicated that a flow rate of 0.45 mL min⁻¹, a temperature of 41.88 °C and a pH of 8.57 would give the best separation. The optimum region (yellow region) in which there was the best separation of each peak is shown in Fig. 8.

Fig. 8 At a flow rate of 0.45 mL min⁻¹, the optimum region of the best separation.

3.3.5. Verification of the optimized conditions and the predictive model. The fitness of the predicting response values were investigated under the conditions of: flow rate 0.45 mL min⁻¹, temperature 41.88 °C and pH 8.57. The experiments performed to verify the validity of the optimized conditions were carried out six times under the same conditions and the RSD% is presented in Table 4. All peaks of interests gave a good separation; the critical resolution was 2.0 or better. It shows that the developed quadratic models were well suited and the optimal values were valid within the specified range of independent variables. Therefore, the developed method was robust against small changes of the chromatographic parameters. Under the stress conditions, the baseline separation of all peaks of the impurity and degradation products and the API indicated the stability-indicating performance of the method.

Table 4 Experimental values of the responses at the optimum conditions

Number	Optimal levels of response parameters
	Rs1	Rs2	Rs3	TIMM
1	5.00	2.98	4.51	0.98
2	4.93	2.98	4.51	0.99
3	5.00	2.98	4.51	0.99
4	4.93	2.91	4.51	0.99
5	5.00	2.98	4.51	0.98
6	5.00	2.98	4.51	0.99
RSD%	0.73%	0.96%	0%	0.52%

---

3.4. Method control

A summary of the chromatographic parameters and the final method are given in Table 5. The method was applied for the determination of the impurity and degradation products of IMM in API and pharmaceutical formulations. A typical chromatogram of IMM and all of its impurity and degradation products (A–E) is shown in Fig. 9.

Table 5 Final assay and the stability-indicating method

Column	Waters Acquity UPLC®CSH C18, 2.1 mm × 50 mm, 1.7 μm
Eluent A	0.1% g mL^−1 ammonium formate buffer (pH 8.57)
Eluent B	Acetonitrile
Gradient program	(T/%B) 0/5; 1/15; 1.1/25; 5/25; 5.1/40; 7/48; 9/48
Run time	9 min
Flow rate	0.45 mL min^−1
Column temperature	41.88 °C
Injection volume	2 μL
Detection UV	269 nm

---

Fig. 9 Experimental chromatogram of a real sample containing IMM and all of its impurity and degradation products (A–E) for conditions at the optimal values.

Under the developed UPLC conditions, the system suitability parameters were evaluated for IMM and all of its impurity and degradation products. The resolution between any two adjacent components was more than 2.0, and the tailing factor for all the components was between 0.95 and 1.25 (Table 6).

Table 6 System suitability parameters

Compound	RT (min)	RRT^a (n = 6)	USP resolution^b (n = 6)	USP tailing factor (n = 6)
a Relative retention time (RRT) was calculated against the retention time (RT) of IMM.b Resolutions were calculated between two adjacent peaks.
Imp-A	1.03	0.16 ± 0.01	—	1.21 ± 0.02
Imp-B	3.07	0.48 ± 0.00	9.60 ± 0.01	1.04 ± 0.01
Imp-C1	3.69	0.58 ± 0.00	5.00 ± 0.02	1.20 ± 0.01
Imp-D	4.40	0.70 ± 0.00	3.63 ± 0.01	1.19 ± 0.03
Imp-C2	4.82	0.76 ± 0.01	2.98 ± 0.03	1.08 ± 0.01
IMM	6.28	1.00 ± 0.01	7.16 ± 0.01	0.99 ± 0.01
Imp-E	6.91	1.10 ± 0.00	4.51 ± 0.00	1.13 ± 0.01

---

A validation study was performed in compliance with the ICH guideline Q2 (R1). The validation data are shown in Table 7. In addition, the robustness study of the developed method was carried out to evaluate the effect of small variation in the chromatographic conditions. Three main factors (flow rate (±0.05 mL min⁻¹), pH (±0.5) and column temperature (±5 °C)) were chosen for this study. From the results of all the experiments of the deliberately altered conditions, it can be seen that a resolution between two adjacent peaks of 2.0 can be reached. Therefore, it indicated that the method is robust against small changes of the chromatographic conditions.

Table 7 Validation study: validation data of the final analytical method

Test	Test details	Results
		Imp-A	Imp-B	Imp-C1	Imp-D	Imp-C2	Imp-E	IMM
Specificity		There are no peaks present in the chromatogram of the sample solvent and eluent at the retention time window of IMM and its known impurities and degradation products
Linearity	5 concentration points in the range from LOQ to 200% (R^2)	0.99959	0.99975	0.99998	0.99961	0.99999	0.99972	1.0
Limit of detection	Calculated from S/N = 3	0.64 μg mL^−1	1.78 μg mL^−1	1.63 μg mL^−1	0.91 μg mL^−1	0.16 μg mL^−1	0.96 μg mL^−1	0.088 μg mL^−1
Limit of quantification	Calculated from S/N = 10	1.93 μg mL^−1	5.41 μg mL^−1	4.93 μg mL^−1	2.74 μg mL^−1	0.50 μg mL^−1	2.91 μg mL^−1	0.27 μg mL^−1
Accuracy	Three different levels (RSD)	1.03%	1.85%	2.40%	2.56%	2.63%	2.70%	0.27%
Precision (repeatability)	n = 6 (RSD) (operator A)	1.46%	1.06%	1.09%	1.96%	0.98%	2.08%	1.08%
Intermediate precision	n = 6 (RSD) (operator B)	1.73%	1.11%	1.37%	1.78%	1.25%	3.09%	1.36%

---

3.5. UPLC-Q-TOF-MS/MS study

In our study, the impurity and degradation products were identified through the optimized UPLC-Q-TOF-MS/MS method. The UPLC-ESI-MS/MS spectra of [M + H]⁺ ions of IMM and its impurity and degradation products obtained under the stress conditions are given in Fig. 10.

Fig. 10 The UPLC-ESI-MS/MS spectra of [M + H]⁺ ions of (a) Imp-A (m/z 235) at 10 eV, (b) Imp-B (m/z 526) at 10 eV, (c) Imp-C₁ (m/z 510) at 10 eV, (d) Imp-D (m/z 278) at 10 eV, (e) Imp-C₂ (m/z 510) at 10 eV, (f) IMM (m/z 494) at 10 eV and (g) Imp-E (m/z 528) at 10 eV.

A total of eleven components (containing IMM and the impurity and degradation products) were identified and characterized using the Q-TOF-MS/MS experiments. The data from the Q-TOF-MS/MS of IMM and the impurity and degradation products are given in Table 8. Under the conditions of stress studies, two acid degradation products were received. The molecular formulas were C₁₃H₁₈N₂O₂ ([M + H]⁺ ion, m/z 235) and C₁₆H₁₅N₅ ([M + H]⁺ ion, m/z 278). Four oxidation degradation products were obtained. The molecular formulas were C₂₉H₃₁N₇O₂ ([M + H]⁺ ion, m/z 510), C₁₆H₁₅N₅ ([M + H]⁺ ion, m/z 278), C₂₉H₃₁N₇O₂ ([M + H]⁺ ion, m/z 510) and C₂₉H₃₁N₇O₃ ([M + H]⁺ ion, m/z 526). Under the conditions of the stress study, IMM also had one impurity which was C₂₉H₃₀N₇OCl ([M + H]⁺ ion, m/z 528). The data of the MS/MS fragment ions of all components shown in Table 8 were used to illustrate the structure of each component which is mentioned in the introduction (Fig. 2).

Table 8 Q-TOF-MS/MS data of IMM and its impurity and degradation products along with their elemental composition and major fragments

Compound	Retention time (min)	Molecular formula	Observed mass m/z	Calculated mass m/z	Error (ppm)	MS/MS fragment ions
Imp-A	1.2	C13H19N2O2^+	235.1504	235.1514	4.2	215, 121, 117, 101
Imp-B	3.4	C29H32N7O3^+	526.2537	526.2551	2.6	394, 378, 277, 247, 222, 194
Imp-C1	4.1	C29H32N7O2^+	510.2592	510.2612	3.9	394, 380, 277, 262, 222, 194
Imp-D	4.9	C16H16N5^+	278.1423	278.1410	−4.6	262, 245, 196, 157
Imp-C2	5.4	C29H32N7O2^+	510.2595	510.2612	3.4	423, 395, 277, 222
IMM	6.7	C29H32N7O^+	494.2645	494.2663	3.5	394, 379, 264, 247, 222, 194
Imp-E	7.4	C29H31ClN7O^+	528.2236	528.2253	3.2	428, 392, 377, 281, 220, 180

---

4. Conclusions

In this paper, a simple and rapid UPLC-Q-TOF-ESI-MS/MS method used for the determination of IMM and its impurity and degradation products was rationally developed. The resulting method gave a good separation of all 7 peaks of the sample of IMM and its impurity and degradation products with a critical resolution (Rs, crit > 2.0) well above the baseline separation, with a run time of under 9 min.^22,23

In our study, an innovative quality-by-design approach was used for the development of a stability-indicating, fast, robust and reliable UPLC method. With the assistance of various modeling software programs (DryLab®2000 plus and Design Expert®8.0.6), a great number of experiments were reduced, and highly influential chromatographic parameters were also optimized. Fully considering the factors of separation, analysis time and robustness, an ideal method was developed eventually. In addition, the method was fully validated in compliance with ICH guidelines. The suitability and specificity of the method for the determination of IMM and its impurity and degradation products which were characterized unambiguously using the optimized UPLC-Q-TOF-MS/MS method have been presented in this article.

Acknowledgements

We are thankful to the School of Pharmaceutical Science and Technology (SPST), TianJin university for support with the instruments.

References

1. A. Nageswari, K. V. S. R. Reddy and K. Mukkanti, J. Pharm. Biomed. Anal., 2012, 66, 109–115.
2. W. J. Szczepek, B. Kosmacińska, A. Bielejewska, W. Łuniewski, M. Skarżyński and D. Rozmarynowska, J. Pharm. Biomed. Anal., 2007, 43, 1682–1691.
3. V. V. Vivekanand, D. Sreenivas Rao, G. Vaidhyanathan, N. M. Sekhar, S. Avijit Kelkar and P. Ramachandra Puranik, J. Pharm. Biomed. Anal., 2003, 33, 879–889.
4. D. Ivanovic, M. Medenica, B. Jancic and A. Malenovic, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2004, 800, 253–258.
5. M. Medenica, B. Jancic, D. Ivanovic and A. Malenovic, J. Chromatogr. A, 2004, 1031, 243–248.
6. ICH Q1A (R2), Stability Testing of New Drug Substances and Products, February, 2003.
7. US Pharmacopoeia Convention, The United States Pharmacopoeia, MD, 26th edn, 2003, pp. 1151–1154.
8. W. Lihong, H. Yuan, Z. Yanqiao, Z. Sujuan and W. Baowei, Chin. J. Pharm., 2011, 42, 728–731.
9. Z. Sujuan, Y. Lixia, C. Xiaolei, S. Qiannan and L. Min, Chin. J. Pharm. Anal., 2013, 33, 633–637.
10. B. Nickerson, Sample Preparation of Pharmaceutical Dosage Forms, Springer, Heidelberg, 2011.
11. ICH Q8 (R2), Pharmaceutical development, August, 2009.
12. FDA, A risk-based approach, MD, Rockville, August, 2002.
13. K. Monks, I. Molnár, H. J. Rieger, B. Bogáti and E. Szabó, J. Chromatogr. A, 2012, 1232, 218–230.
14. A. H. Schmidt and I. Molnár, J. Pharm. Biomed. Anal., 2013, 78–79, 65–74.
15. P. N. Patel, D. Rajesh Kumar, S. Gananadhamu and R. Srinivas, RSC Adv., 2015, 5, 21142–21152.
16. P. D. Kalariya, M. Sharma, P. Garg, J. Reddy Thota, S. Ragampeta and M. V. N. Kumar Talluri, RSC Adv., 2015, 5, 31024–31038.
17. R. M. Borkar, B. Raju, R. Srinivas, P. Patel and S. K. Shetty, Biomed. Chromatogr., 2012, 26, 720–736.
18. V. M. Rodríguez-González, A. Femenia, R. Minjares-Fuentes and R. F. González-Laredo, LWT--Food Sci. Technol., 2012, 47, 225–232.
19. J. Yanping, W. Yunhai, C. Julin and W. Yunying, J. Taiwan Inst. Chem. Eng., 2014, 45, 589–595.
20. S. K. Soni, N. Goyal, J. K. Gupta and R. Soni, Starch - Stärke, 2012, 64, 64–77.
21. G. J. Swamy, A. Sangamithra and V. Chandrasekar, Dyes Pigm., 2014, 111, 64–74.
22. I. Molnár, H. J. Rieger and K. E. Monks, J. Chromatogr. A, 2010, 1217, 3193–3200.
23. R. M. Bianchini, P. M. Castellano and T. S. Kaufman, Anal. Chim. Acta, 2009, 654, 141–147.

---