Accepted Manuscript Application of Quality by Design to optimize a stabilityindicating LC method for the determination of ticagrelor and its impurities
Nathalie Wingert, Jéssica B. Ellwanger, Lívia M. Bueno, Caren Gobetti, Cássia V. Garcia, Martin Steppe, Elfrides E.S. Schapoval PII: DOI: Reference:
S0928-0987(18)30144-1 doi:10.1016/j.ejps.2018.03.029 PHASCI 4458
To appear in:
European Journal of Pharmaceutical Sciences
Received date: Revised date: Accepted date:
4 December 2017 21 February 2018 24 March 2018
Please cite this article as: Nathalie Wingert, Jéssica B. Ellwanger, Lívia M. Bueno, Caren Gobetti, Cássia V. Garcia, Martin Steppe, Elfrides E.S. Schapoval , Application of Quality by Design to optimize a stability-indicating LC method for the determination of ticagrelor and its impurities. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Phasci(2017), doi:10.1016/j.ejps.2018.03.029
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Application of Quality by Design to optimize a stability-indicating LC method for the determination of ticagrelor and its impurities
Nathalie Wingert a*, Jéssica B. Ellwanger a, Lívia M. Bueno a, Caren Gobetti a, Cássia V. Garcia a, Martin Steppe a, Elfrides E. S. Schapoval a
a
Laboratory of Pharmaceutical Quality Control – State University of Rio Grande do Sul, Porto
PT
Alegre, Brazil.
RI
* Correspondent author:
[email protected], phone: +555133085214
SC
ABSTRACT
Simultaneous analysis of drug compounds and their impurities of degradation and synthesis
NU
became constant in the modern pharmaceutical analysis. Likewise, analytical techniques must improve sensitivity and selectivity for the monitoring of pharmaceutical products, allowing a full
MA
assessment of impurities in drug products and, therefore, ensure safety and efficacy of pharmacological treatments. The application of Quality by Design (QbD) principles has proved to be feasible on the elaboration of analytical methods, allowing the comprehensive evaluation and
D
measurement of different analytical parameters and their effects on critical properties of the
PT E
methodology in development. QbD approach was applied to the development of a fast and selective HPLC method for the analysis of the antiplatelet aggregation drug ticagrelor and its degradation products in presence of three impurities of synthesis. Fractional factorial resolution V
CE
was the screening experimental design applied to five method parameters. Response surface methodology was carried by central composite star face design on the two critical method
AC
parameters selected. Analytical design space, established after the application of Monte-Carlo simulations, verified whether predicted results were in accordance with critical quality attributes. The developed and validated HPLC method with DAD detection at 225 nm was able to resolve eight related compounds in less than three minutes.
Keywords: Quality by Design; Drug impurities; Ticagrelor; HPLC; AQbD
ACCEPTED MANUSCRIPT 1. Introduction Quality by design (QbD) concepts were initially applied to the design of product and process, especially with the release of ICH Q8 guide on 2008 (ICH, 2009). On the last few years, QbD has been successfully employed in the development and optimization of analytical methods and, through tools like risk assessment, improving the knowledge and reliability of accurate methodologies for monitoring samples with complex matrices or multiple analytes (Hubert et al., 2014; Monks et al., 2012; Orlandini et al., 2013; Vogt and Kord, 2010). Such application has
PT
emerged the term analytical quality by design (AQbD) (Hubert et al., 2015; Reid et al., 2013). The most representative examples of the application of AQbD concepts are on the analysis of active
RI
pharmaceutical compounds, especially in presence of its impurities and degradation products (Furlanetto et al., 2013; Karmarkar et al., 2014; Terzić et al., 2016; Tol et al., 2016). Through
SC
experimentally designed tests, method parameters with critical influence on the analytical methodology are determined. The critical evaluation where analytical characteristics can be
NU
modulated through changes in method parameters is called risk assessment (ICH, 2009; Orlandini et al., 2013). The implementation of AQbD and proper risk assessment along method development
MA
requires a prior understanding of analytical variables to ensure the creation of reproducible procedures with effective strategic control, thus reducing the number of analysis and
D
development time. System suitability factors are evaluated and classified according to their
PT E
representation of an optimized analytical system. Owning such information, application of experimental design with appropriate software starts the modelling process to develop a multidimensional space; where it is possible to verify which parameters and factors under analysis
CE
have greater or reduced significance for the analytical method (Hubert et al., 2015; Rozet et al., 2013). At the end, the developed system will be presented with defined experimental spaces,
AC
allowing analytical flexibility with attested quality. Once the complete elimination of impurities from pharmaceutical products is not factual, the analysis and report of drugs together with their impurities of degradation and synthesis became constant in the modern pharmaceutical analysis. Furthermore, analytical techniques with improved sensitivity and selectivity allowed a better and faster monitoring of pharmaceutical products, assigning the full control of impurities in drug products and ensuring an effective therapy (Jain and Basniwal, 2013; Singh et al., 2013). Ticagrelor (TIC) is an inhibitor of platelet aggregation (Ticagrelor., 2010) and, likewise several pharmaceutical drug products prescribed for continued use in patients with unstable health conditions, it requires the monitoring of pharmaceutical impurities to ensure safety and efficacy of pharmacological treatments (ICH, 2006a, 2006b). As defined by ICH guideline Q3A (R2), starting materials, intermediates, and
ACCEPTED MANUSCRIPT degradation products are classified as organic impurities and have specific limits for report and quantification according to pharmaceutical dosage (ICH, 2006b). TIC presents different synthesis pathways and consequently several organic impurities that can emerge during the manufacturing process. Impurities T1A ((1R,2S)-2-(3,4-difluorophenyl)cyclopropanamine); T1B (2-hydroxy2phenylacetic acid) and T2 (4,6-dichloro-2(propylthio)pyrimidin-5-amine) (Figure 1) are starting materials present on two alternative processes of TIC synthesis (Springthorpe et al., 2007; Zhang et al., 2012).
PT
Current scientific literature presents the analysis of TIC as bulk and pharmaceutical product on high-performance liquid chromatography (HPLC) and ultraviolet spectroscopy (UV) (Gobetti et
RI
al., 2015; Oliveira et al., 2016). Other recent works have presented LC analysis of TIC with the identification of degradation products (DPs) and impurities (Bueno et al., 2017; Kumar et al., 2016;
SC
Yaye et al., 2015). Despite the analytical methods reported in renewed journals, none of them accomplished the concurrent evaluation of TIC, synthesis impurities and DPs all together.
NU
Therefore, AQbD approach and workflow were applied to the development and optimization of a fast and selective HPLC method for the analysis of TIC (commercial product and bulk) and its DPs
PT E
[FIGURE 1]
D
MA
in presence of organic impurities T1A, T1B, and T2.
CE
2. Methodology
2.1. Chemicals and reagents
AC
TIC standard chemical substance (99.7%), impurity T1 supplied as T1A:T1B (1:1) (99.3%) and impurity T2 (99.9%) were purchased from Sequoia Research Products (Pangbourne, UK). TIC 90 mg tablets (Brilinta®, AstraZeneca) were acquired on local market. Acetonitrile (ACN) was the gradient grade for liquid chromatography and further reagents from analytical class.
2.2. Solutions and sample preparation Stock solutions of TIC, impurities T1 and T2 were prepared independently by weighting an accurate amount and dissolving it in ACN, obtaining stock solutions of 200 µg/mL. Ten tablets of TIC were grounded and mixed into a homogenous powder; an amount correspondent to 10 mg of TIC was weighed and dissolved in 20 mL of ACN. After 15 min of sample sonication to improve solubility, resultant dispersion of 500 µg/mL was filtered with a quantitative paper filter (8 µm
ACCEPTED MANUSCRIPT pore size). All stock solutions were individually stored in glass flasks protected from light at 8 °C. Work samples were daily prepared by diluting stock solutions with mobile phase up to desired final concentration. In order to evaluate the rise of DPs, TIC sample and standard substance solutions were exposed to the following stress conditions: photodegradation was tested exposing TIC solution (200 µg/mL) to UVC radiation of 27000 W.h/m2 on a mirrored chamber with 0.03 m3; 0.03% hydrogen peroxide was mixed with drug solution at 200 µg/mL for oxidative degradation;
PT
hydrolysis was verified exposing TIC 200 µg/mL to acidic (1M HCl) and alkaline (1M NaOH) solutions. All stress conditions were held up to 5 hours at 25 °C. Hydrolysis solutions were
RI
neutralized and TIC final concentration was 66.67 µg/mL. Photolysis and oxidative sample solutions were diluted with mobile phase to the final theoretical concentration of 30 µg/mL.
SC
Along methodology optimization, the general composition of mobile phase was 25-75 mM
2.3. Apparatus and working conditions
MA
filtered before use on HPLC on 0.45 µm filters.
NU
ammonium acetate, pH 8.1-8.5, and ACN varying from 40 to 60%. All samples and solutions were
Analyses were performed on an Agilent HPLC 1200 system (Agilent Technologies, USA) equipped with photodiode array detector. Results were processed by LC solutions software
D
(Agilent Technologies, USA). Compounds separation was tested on C18 Pursuit XR (Varian, Agilent,
PT E
The Netherlands) and C8 Luna (Phenomenex, USA) chromatographic columns sized 50x3 mm and with 3 µm diameters silica particles.
Final LC working conditions, developed after method design were 75 mM ammonium
CE
acetate pH 8.25:ACN (45:55), 0.95 mL/min flow and 30 °C temperature. Separations were
AC
performed on a C18 50x3 mm, 3 µm column with DAD at 225 nm.
2.4. Analytical Quality by Design methodology Design of experiments and statistical analyses were performed on MODDE® 13 software (Sartorius Stedim Biotech, Umetrics, Sweden). QbD workflow followed a sequence of events based on ICH Q8 (ICH, 2009). The concepts and work structure presented on recent AQbD applications were starting points for method development and walked through in details along the manuscript (Furlanetto et al., 2013; Orlandini et al., 2013; Parr and Schmidt, 2017; Peraman et al., 2015). Analytical target profile (ATP) was initially set in order to guide the list of critical quality attributes (CQAs), considering the analytical methodology to be applied. CQAs describe the desired conditions on the analysis to
ACCEPTED MANUSCRIPT guarantee the ATP. Critical method parameters (CMPs) are connected to CQAs through the risk assessment. Experiments were divided into two moments, screening and optimization. Adequate design of experiment (DoE) was performed at each stage, resulting in statistical data regarding the effect of evaluated CMPs on CQAs. Appropriate analysis of effects resulted in knowledge space and design space. Final data simulation was performed in order to assess whether the designed method could comply with CQAs specifications. Method validation is fundamental to guarantee a reliable method for routine analysis.
PT
After determination of analytical method design space (DS) and robust region, ICH Q3A (R2), 2006 and Q2(R1), 2005 guidelines were used to orientate HPLC method validation for TIC sample in
RI
presence of impurities T1A, T1B and T2 (ICH, 2006a, 2005).
SC
3. Results and discussion 3.1. Preliminary studies
NU
In order to recognize the analytical variables with greater importance on the separation of TIC, impurities T1A, T1B, T2, and DPs, initial data were acquired following previously developed
MA
HPLC methods (Bueno et al., 2017; Gobetti et al., 2015) with adaptations to smaller columns (5 cm long). Starting from preliminary results, it was possible to gather information related to the chromatographic behaviour of compounds to be analysed. Although TIC susceptibility to stressed
D
media has already been reported (Bueno et al., 2017; Kumar et al., 2016; Yaye et al., 2015), a new
PT E
study was conducted in order to standardize the experimental conditions and to evaluate the main degradation products formed in each condition. One important feature verified during initial LC analyses was the necessity of working with
CE
mobile phase at a pH higher than 8.0. Imp T1A (predicted pKa 8.07) was kept in its non-ionized form, increasing the interaction with the stationary phase and consequently, the selectivity of
AC
peaks from Imps T1A from T1B.
Based on formerly developed methods, C8 chromatographic column was initially applied. Later tests with a C18 column, showed an improvement in peak shape and resolution. Therefore, C18 HPLC column with 3 µm particles and dimensions of 3x50 mm was chosen as separation instrument. Stability tests performed on TIC stock sample solutions resulted in the formation of two main DPs after 2 h of exposition to UVC radiation; one DP formed due acid hydrolysis after 5 h, and another DP was originated from oxidative degradation with 2 h of exposition to 0.03% H 2O2. Stress conditions were not held for a too long time in order to avoid the formation of secondary DPs. Initial analyses of mixed compounds showed poor resolution between Imp T1A and TIC. After
ACCEPTED MANUSCRIPT TIC stress study, four peaks from DPs were detected, DPAC (from acid hydrolysis) and DPUVC1 (from UVC radiation) enclosed by Imp T1A and TIC peaks, and also DPUVC2 and DPOX (from oxidative degradation) peaks between TIC and Imp T2. One of the peaks detected with degradation protocol, DPuvc2, has been previously elucidated as a product of TIC loss of (3,4difluorophenyl)cyclopropyl moiety (Bueno et al., 2017; Kumar et al., 2016; Yaye et al., 2015). The accurate analysis of TIC stressed samples in presence of synthesis impurities demanded an increase in peak capacity, so LC method would be able to consistently analyse eight compounds
PT
simultaneously.
RI
3.2. Method development and optimization
The application of AQbD systematic workflow initiates with a consistent definition of the
SC
main objectives of the method to be established. ATP was defined as the development of a simple, fast, stability-indicating HPLC method to assess TIC in the pharmaceutical product along with three
NU
impurities of synthesis and four DPs. Parr and Schmidt 2017 proposed that ATP should be defined before method development and just then connected to the method goals (Parr and Schmidt,
MA
2017); however, it is necessary to differentiate between specifications and characteristics of analytical methodologies. This can be achieved through risk assessment and the statement of CQAs, which will define method scouting phase on the search for the adequate analytical method.
D
CQAs were initially described based on previous knowledge of drug analysis by LC and early
PT E
assessment of TIC and impurities samples. Factors considered as CQA are peak resolution (Res); analysis time; retention factor (k), and the maximum pressure of the system. Quality risk assessment can be defined as the joint of CQAs and CMPs, supporting the
CE
evaluation of the degree and magnitude each CMPs affects CQAs, and how they can be optimized (Furlanetto et al., 2013; ICH Expert Working Group, 2005). This approach was applied through all
AC
method development stages, ensuring a comprehensive view of the analytical process. Ishikawa diagram was the tool chosen to express risk assessment (Figure 2), focusing on the different factors with an influence on the analytical method. Parameters related to LC system (temperature and flow) and mobile phase (salt concentration, the percentage of ACN and pH) were selected for investigation as CMPs. DoE was applied to investigate the parameters with critical influence on analytical methodology in two stages: screening and optimization. Later on, CQAs were assessed and classified according to their response to an adequate analytical system, according to ATP.
[Figure 2]
ACCEPTED MANUSCRIPT
3.3. Screening phase Compared to traditional method development, altering one factor at a time, application of DoE accelerates the search for variables with greater influence on CQAs. AQbD screening phase comprises the evaluation of interactions and changes on CMPs through a fitted DoE with, analysis of main effects on CQAs. Plackett-Burnman, complete and fractional factorial, D-Optimal, and
PT
Rechtschaffner are the most used linear and interaction DoE for screening purpose (Hibbert, 2012). Considering all response variables (Tables 1 and 2), the experimental design chosen was
RI
the Fractional Factorial Resolution V, with 16 experiments at two levels and three central points
NU
factors interactions free of alias (Montgomery, 2001).
PT E
D
MA
[Table 1]
[Table 2]
SC
(α=0.05), offering a mathematical description of the system behaviour with main effects and two
To better understand the investigation of knowledge space, contour plots for analysis time
CE
and k (Figure 3) and the standardized effects of linear and interaction factors on T1B-T1A, T1A-TIC, and TIC-T2 peak resolution (Figure 4) are displayed. Linear terms ACN and Flow presented a
AC
significant negative effect on T1B-T1A resolution and TIC-T2 resolution. However, the major effect detect was related to the reduction of T1B-T1A resolution and T1A-TIC resolution with the increase of Ammonium Acetate (AmmAcet) concentration. Since greater values of k for T1B were found when AmmAcet was in its higher level (Figure 3A), this factor was fixed at 75 mM for the next DoE. As specified before, pH of mobile phase has a minimum value starting from 8.0. Once pH effect on peak resolutions was both positive and negative, it was kept at its mean level for optimization. The minimal target value for k of the first detected peak (Imp T1B) was selected due to the high polarity of the molecule, which had reduced interaction to the stationary phase. System pressure had significant changes due linear terms flow, ACN and temperature. However, all obtained results were between 77 and 145 bar, quite under the selected limit of 200 bar,
ACCEPTED MANUSCRIPT therefore pressure was no longer consider a CQA. Fractional factorial Res V was found to be statistically significant and with no lack of fit (P>0.05) for all responses evaluated on the design.
PT
[Figure 3]
SC
RI
[Figure 4]
NU
3.4. Optimization phase
After screening phase, CQAs were reviewed and refined through risk assessment. ACN and
MA
flow were the variables with greater impact on CQAs and, therefore, must be controlled and closely evaluated in order to obtain the desired quality of methodology. ACN (47 to 57%) and flow (0.8 to 1.0 mL/min) CMPs combinations were studied through
D
the response surface method Central Composite Face (CCF), with star distance 1, composed of 8
PT E
experiments and 3 central points (α=0.05). CQAs were revised, reducing the desired analysis time from 4.0 to 3.5 min. Resolution between peaks was kept as critical response and calculated accordingly.
CE
In order to assure the application of developed method as stability indicating for TIC drug product, minimum and target resolution values were recalculated considering the necessity of
AC
detect peaks of two DPs (DPAC and DPUVC1) between Imp T1A and TIC peak, as well as another two peaks from DPs (DPOXI and DPUVC2) in the time range from TIC to Imp T2 peaks. The mathematical equation for peak resolution (Dong, 2006; Kazakevich and LoBrutto, 2007) was applied with the purpose of establishing the minimum resolution values for T1A-TIC resolution and TIC-T2 resolution. Once four peaks were detected in each case, the minimum resolution was set as 1.5 and target as 2.0 for each couple of peaks (considering TIC, impurities, and DPs). The mean values for peak width of each substance were obtained from previous analysis and applied on the mathematical equations. Updated threshold values for method CQAs are described in Table 3.
[Table 3]
ACCEPTED MANUSCRIPT As presented on resultant contours plots (Figure 5). Values set as targets for Time, ResT1BT1A and ResT1A-TIC are within the response surface assessed through CCF design. Although k for T1B never reached the target value, it was registered over the minimum threshold on most of the experiments. The optimum values for each response and the statistical evaluation of CCF model
PT
are presented in Table 4.
SC
RI
[Figure 5]
Values of coefficients R2, R2 Adjust, and R2 Prediction represent the acceptability of CCF
NU
model for each CQA. As can be noticed by the lower value of R2 Prediction for k (T1B), CCF could not adequately model the variations in this response. Such lack of adjusts and predictability
MA
occurred due to the reduced effect of CMPs on Imp T1B retention time, and consequently on k.
PT E
D
[Table 4]
Once moving within the DS is not considered a significant change in the method (ICH, 2009) and the confidence of responses inside the designed region was proved, all area in the DS can be
CE
chosen as a working point. Comprising CQAs limits and including target values for all responses, optimization was performed to obtain ideal method set point. Randomly distributed results of
AC
CCF design were assessed by reliability analysis and the probability of failure acquired by Monte Carlo simulation. The estimation of failure probability allowed the discretization of experimental region by means of the possibility of not achieving the desired CQAs thresholds, resulting on the DS with hypersurface limiting trusted region from failure area. The resultant DS created with AQbD approach had hypercube edges at 0.89 and 1 mL/min for mobile phase flow and 49 and 53% for ACN percentage on mobile phase, define the method region with 1% risk of failure (Figure 6).
[Figure 6]
ACCEPTED MANUSCRIPT The developed method presented a defined experimental space, allowing analytical flexibility with attested quality (Figure 6). Final method parameters were 75 mM ammonium acetate:ACN (49:51), pH 8.3, 0.95 mL/min flow and 32.5 °C temperature.
PT
[Figure 7]
RI
3.5. Working points and robustness for method control
Following the sequential steps on AQbD approach, after method optimization and
SC
establishment of design space, it is recommended to create strategies for method control in order to keep all designed parameters working according to selected CQAs. Method verification was
NU
performed through robustness assessment by evaluation of TIC commercial samples at 50 µg/mL, spiked with impurities T1A, T1B, and T2 at 5 µg/mL each. Plackett-Burman experimental design
MA
was applied to routine method central points with small and controlled variations, in order to confirm whether those changes affect CQAs significantly. Changes were performed on ACN (49 and 53%), flow (0.9 and 1.0 mL/min), pH (8.2 and 8.5), and temperature (30 and 35 °C),
D
randomized into 8 runs with 3 central points. Drug quantification and monitoring of system
PT E
suitability parameters were performed to certify method ability on achieving all performance requirements. None of the changes implemented showed a significant variation in TIC assay, peak separation, or analysis time. The mean values found along robustness assessment to peak
CE
resolution were ResT1B-T1A: 2.26, ResT1A-TIC: 4.21, ResTIC-T2: 9.49. Analysis time ranged from 2.36 to 2.89 min. The clear definition of performance characteristics assists routine analysis,
AC
assuring maintenance of quality settings developed and verified through risk assessment.
3.6. Method validation
After the development of HPLC method with the flexible experimental region and robust results, validation was performed according to specifications of international guidelines (ICH, 2005). Linearity range was established so impurities T1 (A and B) and T2 could be detected even at levels lower than the identification threshold set out for TIC samples (Table 5). Limits of detection (LoD) and quantification (LoQ) were set according to signal-to-noise rate. The concentrations found for each intensity were experimentally tested in order to find accurate concentration values for TIC and both impurities (Table 5).
ACCEPTED MANUSCRIPT [Table 5]
Accuracy for TIC and impurities in pharmaceutical formulation was assessed in three concentrations, respectively, low (15 µg/mL and 1.5 µg/mL), mean (30 µg/mL and 3 µg/mL), and high (40 µg/mL and 4 µg/mL). Recoveries of TIC standard and impurities are shown in Table 5.
PT
Precision and accuracy results are in agreement with values proposed by Barnett et al. for drug quantification together with impurities and excipients, 100% ±3% for accuracy and ≤1% for
RI
precision (Barnett et al., 2016).
Responses found for method linearity, intraday and interday precision, LoD, LoQ (Table 5)
SC
and accuracy were within the desired ranges to assure analytical reliability.
NU
4. Conclusion
Modern technologies related to the analysis of multiple compounds demand methods
MA
developed in a more comprehensive way. AQbD approach allows the evaluation and measurement of different analytical parameters and their effects on critical properties of the methodology. The application of such procedure reinforces the importance of scientific knowledge
D
over routine pharmaceutical analysis by the possibilities of improvement when searching for
PT E
optimized methodologies. Moreover, the accurate analysis of synthesis impurities and DPs, as required by international regulation entities, demands the increment of refinement on analytical methodology development.
CE
According to ATP requirements, a fast and reliable HPLC method was developed and validated for the analysis of TIC in drug products along with three impurities of synthesis and four
AC
DPs. Chromatographic analysis was performed up to 2.8 minutes and resolved the eight compounds with a resolution larger than 2.0. Data processing by statistical software allowed the precise measurement of %ACN and flow effects, combined or individually, on selected CQAs responses. Method operable design region was found after experimental design (orthogonal and response surface) and statistical considerations. Besides optimum response and minimum values, AQbD involves the continuous verification of method performance, which was implemented through application of robustness assessment and evaluation of method system suitability.
References
ACCEPTED MANUSCRIPT
AC
CE
PT E
D
MA
NU
SC
RI
PT
Barnett, K.L., McGregor, P.L., Martin, G.P., LeBlond, D.J., Weitzel, M.L.J., Ermer, J., Walfish, S., Nethercote, P., Gratzl, G.S., Kovacs, E., 2016. Stimuli article: Analytical Target Profile and application throughout the analytical lifecycle. Pharmacopeial Forum. Bueno, L.M., Manoel, J.W., Giordani, C.F.A., Mendez, A.S.L., Volpato, N.M., Schapoval, E.E.S., Steppe, M., Garcia, C.V., 2017. HPLC method for simultaneous analysis of ticagrelor and its organic impurities and identification of two major photodegradation products. Eur. J. Pharm. Sci. 97, 22–29. https://doi.org/10.1016/j.ejps.2016.11.004 Dong, M.W., 2006. Modern HPLC for Practicing Scientists. John Wiley & Sons, Inc. https://doi.org/10.1002/0471973106 Furlanetto, S., Orlandini, S., Pasquini, B., Del Bubba, M., Pinzauti, S., 2013. Quality by Design approach in the development of a solvent-modified micellar electrokinetic chromatography method: Finding the design space for the determination of amitriptyline and its impurities. Anal. Chim. Acta 802, 113–124. https://doi.org/10.1016/j.aca.2013.10.005 Gobetti, C., Pereira, R.L., Mendez, A.S.L., Garcia, C.V., 2015. Determination of the New Antiplatelet Agent Ticagrelor in Tablets by Stability-Indicating HPLC Method. Curr. Pharm. Anal. 10, 279– 283. Hibbert, D.B., 2012. Experimental design in chromatography: A tutorial review. J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 910, 2–13. https://doi.org/10.1016/j.jchromb.2012.01.020 Hubert, C., Houari, S., Rozet, E., Lebrun, P., Hubert, P., 2015. Towards a full integration of optimization and validation phases: An analytical-quality-by-design approach. J. Chromatogr. A 1395, 88–98. https://doi.org/10.1016/j.chroma.2015.03.059 Hubert, C., Lebrun, P., Houari, S., Ziemons, E., Rozet, E., Hubert, P., 2014. Improvement of a stability-indicating method by Quality-by-Design versus Quality-by-Testing: A case of a learning process. J. Pharm. Biomed. Anal. 88, 401–409. https://doi.org/10.1016/j.jpba.2013.09.026 ICH, 2009. International Conference on Harmonisation - Pharmaceutical Development Q8(R2). ICH, 2006a. International Conference on Harmonisation - Impurities in new drug products Q3B(R2). ICH, 2006b. International Conference on Harmonisation - Requirements for registration of pharmaceuticals Q3a(R2). ICH, 2005. International Conference on Harmonisation - Validation of analytical procedures: text and methodology Q2(R1). ICH Expert Working Group, 2005. Quality Risk Management Q9. ICH Harmon. Tripart. Guidel. 1–23. https://doi.org/10.1007/s11095-007-9511-1 Jain, D., Basniwal, P.K., 2013. Forced degradation and impurity profiling: Recent trends in analytical perspectives. J. Pharm. Biomed. Anal. 86, 11–35. https://doi.org/10.1016/j.jpba.2013.07.013 Karmarkar, S., Yang, X., Garber, R., Szajkovics, A., Koberda, M., 2014. Analysis Quality by design ( QbD ) based development and validation of an HPLC method for amiodarone hydrochloride and its impurities in the drug substance. J. Pharm. Biomed. 100, 167–174. Kazakevich, Y., LoBrutto, R., 2007. HPLC for Pharmaceutical Scientists. John Wiley & Sons, Inc. https://doi.org/10.1002/9780470087954 Kumar, N., Devineni, S.R., Gajjala, P.R., Gupta, D.K., Bhat, S., Kumar, R., Dubey, S.K., Kumar, P., 2016. Four process-related potential new impurities in ticagrelor: Identification, isolation, characterization using HPLC, LC/ESI-MSn, NMR and their synthesis. J. Pharm. Biomed. Anal. 120, 248–260. https://doi.org/10.1016/j.jpba.2015.12.037 Monks, K., Molnár, I., Rieger, H.J., Bogáti, B., Szabó, E., 2012. Quality by Design: Multidimensional exploration of the design space in high performance liquid chromatography method development for better robustness before validation. J. Chromatogr. A 1232, 218–230. https://doi.org/10.1016/j.chroma.2011.12.041 Montgomery, D.C., 2001. Design and Analysis of Experiments, 5th ed. John Wiley & Sons.
ACCEPTED MANUSCRIPT
AC
CE
PT E
D
MA
NU
SC
RI
PT
https://doi.org/10.1198/tech.2006.s372 Oliveira, S., Silva Bitencourt, A., Gobetti, C., Loureiro Mendez, A., Garcia, C., 2016. Assaying the antiplatelet ticagrelor by validated UV spectrophotometric method with performance equivalent to HPLC. Curr. Pharm. Anal. 12, 1–1. https://doi.org/10.2174/1573412912666160901121203 Orlandini, S., Pinzauti, S., Furlanetto, S., 2013. Application of quality by design to the development of analytical separation methods. Anal. Bioanal. Chem. 405, 443–450. https://doi.org/10.1007/s00216-012-6302-2 Parr, M.K., Schmidt, A.H., 2017. Life cycle management of analytical methods. J. Pharm. Biomed. Anal. 1–12. https://doi.org/10.1016/j.jpba.2017.06.020 Peraman, R., Bhadraya, K., Padmanabha Reddy, Y., 2015. Analytical Quality by Design: A Tool for Regulatory Flexibility and Robust Analytics. Int. J. Anal. Chem. 2015, 1–9. https://doi.org/10.1155/2015/868727 Reid, G., Morgado, J., Harrington, B., Wang, J., Fortin, D., 2013. Analytical Quality by Design (AQbD) in Pharmaceutical Development. Am. Pharm. Rev. 1–15. Rozet, E., Lebrun, P., Debrus, B., Boulanger, B., Hubert, P., 2013. Design Spaces for analytical methods. Trends Anal. Chem. 42, 157–167. Singh, S., Junwal, M., Modhe, G., Tiwari, H., Kurmi, M., Parashar, N., Sidduri, P., 2013. Forced degradation studies to assess the stability of drugs and products. Trends Anal. Chem. 49, 71– 88. https://doi.org/10.1016/j.trac.2013.05.006 Springthorpe, B., Bailey, A., Barton, P., Birkinshaw, T.N., Bonnert, R. V., Brown, R.C., Chapman, D., Dixon, J., Guile, S.D., Humphries, R.G., Hunt, S.F., Ince, F., Ingall, A.H., Kirk, I.P., Leeson, P.D., Leff, P., Lewis, R.J., Martin, B.P., McGinnity, D.F., Mortimore, M.P., Paine, S.W., Pairaudeau, G., Patel, A., Rigby, A.J., Riley, R.J., Teobald, B.J., Tomlinson, W., Webborn, P.J.H., Willis, P.A., 2007. From ATP to AZD6140: The discovery of an orally active reversible P2Y12 receptor antagonist for the prevention of thrombosis. Bioorganic Med. Chem. Lett. 17, 6013–6018. https://doi.org/10.1016/j.bmcl.2007.07.057 Terzić, J., Popović, I., Stajić, A., Tumpa, A., Jančić-Stojanovic, B., 2016. Application of Analytical Quality by Design concept for bilastine and its degradation impurities determination by hydrophilic interaction liquid chromatographic method. J. Pharm. Biomed. Anal. 125, 385– 393. https://doi.org/10.1016/j.jpba.2016.04.022 Ticagrelor., 2010. EMA: SUMMARY OF PRODUCT CHARACTERISTICS. [WWW Document]. URL http://www.ema.europa.eu/docs/en_GB/document_library/EPAR__Product_Information/human/001241/WC500100494.pdf (accessed 7.27.17). Tol, T., Kadam, N., Raotole, N., Desai, A., Samanta, G., 2016. A simultaneous determination of related substances by high performance liquid chromatography in a drug product using quality by design approach 1432, 26–38. https://doi.org/10.1016/j.chroma.2015.12.080 Vogt, F., Kord, A., 2010. Development of Quality-by-Design Analytical Methods. J. Pharm. Sci. 3, 26–33. https://doi.org/10.1002/jps Yaye, H.S., Secrétan, P.H., Henriet, T., Bernard, M., Amrani, F., Akrout, W., Tilleul, P., Yagoubi, N., Do, B., 2015. Identification of the major degradation pathways of ticagrelor. J. Pharm. Biomed. Anal. 105, 74–83. https://doi.org/10.1016/j.jpba.2014.11.046 Zhang, H., Liu, J., Zhang, L., Kong, L., Yao, H., Sun, H., 2012. Synthesis and biological evaluation of ticagrelor derivatives as novel antiplatelet agents. Bioorg. Med. Chem. Lett. 22, 3598–602. https://doi.org/10.1016/j.bmcl.2012.04.050
ACCEPTED MANUSCRIPT Figure 1. Chemical structures of ticagrelor, impurity T1A and B and impurity T2.
Figure 2. Analytical parameters highlighted on initial risk assessment, those in bold were chosen for screening DoE.
Figure 3. Contour plots knowledge space evaluation. A) Results for k of Imp T1B by plotting ACN vs. temperature at concentrations of 25, 50 and 75 mM Ammonium Acetate. B) Analysis time
PT
results with flow vs. ACN at three temperatures: 25, 32.5, and 40 °C. Fixed factors: 50 mM
RI
Ammonium Acetate and pH 8.3.
SC
Figure 4. Effect plots for peak resolution after Fractional Factorial experiment. Resolution between peaks from T1B-T1A (red, -----); T1A-TIC (green, -·-·-·); TIC-T2 (blue, ·····). A: AmmAcet; B: ACN; C:
NU
ACN*AmmAcet; D: ACN*Flow; E: ACN*pH; F: ACN*Temp; G: Flow*AmmAcet; H: Flow; I: pH; J: pH*AmmAcet; K: pH*Flow; L: pH*Temp; M: Temp*AmmAcet; N:Temp*Flow; O: Temp. Variables with significant effect on responses have bars exceeding confidence interval limits, represented by
MA
dashed lines.
D
Figure 5. Contour plot for last selected CQAs. Each condition is showed with evaluated ranges of
PT E
CMPs flow vs. ACN.
Figure 6. Design space region with indication of probability of failure for designed CQAs.
CE
Hypercube space generated after Monte Carlo simulations.
AC
Figure 7. Chromatogram of stressed samples of TIC drug product along impurities of synthesis. HPLC parameters at optimum methods conditions after application of AQbD (75 mM ammonium acetate:ACN (49:51)), pH 8.3, 0.95 mL/min, 32.5 °C, and C18 column of (3x50 mm, 3 µm).
ACCEPTED MANUSCRIPT Table 1. Critical method parameters selected after risk analysis and interval of evaluation. Low 45 8.1 25 0.8 25
High 55 8.5 40 1.2 75
AC
CE
PT E
D
MA
NU
SC
RI
PT
Factors Acetonitrile (%) pH Temperature (°C) Flow (mL/min) Ammonium acetate (mM)
ACCEPTED MANUSCRIPT Table 2. Critical quality attributes and performance range applied on screening DoE
Parameters
Minimum
Maximum
Analysis time (min)
-
4
Resolution (T1B-T1A; T1A-TIC; TIC-T2) k for T1B* Pressure (bar)
1.5 0.5 -
200
AC
CE
PT E
D
MA
NU
SC
RI
PT
*k: retention factor.
ACCEPTED MANUSCRIPT Table 3. Threshold and target values for CQAs under evaluation by CCF design. Min 1.5 3.54 3.84 0.5
Target 3.0 2.0 4.67 5.17 0.9
Max 3.5 -
AC
CE
PT E
D
MA
NU
SC
RI
PT
Response Time (min) T1B-T1A resolution T1A-TIC resolution TIC-T2 resolution k for T1B
ACCEPTED MANUSCRIPT Table 4. CQA factors at optimum method condition. Coefficients R2, R2 Adjust (Adj) and R2 Prediction (Pred) obtained with mathematical analysis of CCF model. Optimum
R2
R2 Adj
R2 Pred
Time
2.70
1.000
1.000
0.998
ResT1B-T1A
2.20
0.974
0.948
0.748
ResT1A-TIC
3.89
1.000
0.999
0.998
ResTIC-T2
10.04
0.996
0.992
0.9967
k
0.60
0.832
0.664
-0.449
AC
CE
PT E
D
MA
NU
SC
RI
PT
Response
ACCEPTED MANUSCRIPT Table 5. Method validation results for TIC, and impurities of synthesis T1A, T1B, and T2.
day 1 day 2
Intraday RSD (%) Interday RSD (%, n=12)
Low Mean High
Accuracy (% recovery )
Imp T1A 0.5 - 6.0 0.9993 0.61 0.34 0.69 0.52 103.87 102.92 101.78 0.05 0.20
Imp T1B 0.5 – 6.0 0.9835 0.16 3.02 3.78 2.87 104.52 101.81 102.53 0.15 0.35
AC
CE
PT E
D
MA
NU
SC
RI
LoD (µg/mL) LoQ (µg/mL) *RSD: relative standard deviation.
TIC 5.0 - 60.0 1.0000 0.04 0.52 0.63 0.65 99.62 99.25 99.27 0.08 0.15
PT
Linearity range (µg/mL) R2 Standard error (%)
Imp T2 0.5 - 6.0 0.9991 0.12 0.67 0.31 0.56 97.85 99.10 97.73 0.02 0.08
Graphics Abstract
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7