The development of a new drug product in pharmaceutical industry consists of many processes, including analytical methods. The results obtained by using analytical methods allow us to determine the strategy for further development or provide information on whether the drug product may be released. It is therefore essential that analytical methods are precise, accurate and reliable. High performance liquid chromatography (HPLC) method is the key analytical method for drug product analysis and release. The development of HPLC method may be a challenge because of the complexity of the samples in development. Using the traditional approach (OFAT) the process of HPLC method development and optimization is time consuming and non-transparent, as well as it is impossible to determine factor interactions that affect method performance. In 2004, the FDA first introduced Quality by Design (QbD), the purpose of which is to improve the quality of the drug product by designing it directly into the pharmaceutical process. In June 2018, ICH announced a new ICH Q14 guideline that will include the use of QbD for analytical methods, called Analytical Quality by Design (AQbD). In the introduction part, as a review scientific article, we introduced the theoretical background of AQbD and presented in detail the recent cases, which systematically show the methodology of the AQbD approach to the development and optimization of HPLC methods. In chapter one of experimental part, we focused on the HPLC method from the EP for the determination of related substances in celecoxib. Using the prescribed analytical method the system suitability test (SST) criteria cannot be met: inadequate resolution between celecoxib and impurity B, barely adequate resolution between impurity A and celecoxib. Using the Design of Experiments (DoE) we optimized the pharmacopeial method within the acceptable limits prescribed in the EP. Four critical method parameters were varied using DoE: the ratio of methanol and acetonitrile in the mobile phase, column temperature and mobile phase flow rate. We determined the sweet spot, in which both resolution meet the SST criteria. However, the sweet spot was narrow and the analysis time had to be increased for a factor of 1.5. In chapter two, we started with the development and optimization of a new HPLC method for celecoxib, since the pharmacopeial optimized method was too long and non-robust. In the USP forum, we found the HPLC method for celecoxib capsules, that can be used for the determination of six process related impurities but not also impurity A, as it was not separated from the peak of celecoxib. Our goal was to develop one HPLC method able to determine all seven process related impurities of celecoxib from EP and USP. Due to structural similarity of some investigated process-related impurities (positional isomers), we had to find a more suitable stationary phase for their separation. Using a chiral column in reversed phase, we achieved satisfactory separations. In order to optimize the method and determine the design space, we performed 17 experiments according to a central composite face design, in which the values of three CMPs (the ratio of acetonitrile in the mobile phase, column temperature and mobile phase flow rate) were varied. Three critical resolutions and the retention time of the last eluting impurity were monitored as CMAs. Using the multiple linear regression (MLR) method, we established a statistically significant mathematical model with excellent prediction abilities. Using the Monte Carlo simulations method, we determined the design space, in which all responses meet the criteria with 99 % probability. We also determined a combination of factors that give the optimal response and confirmed the suitable precision, accuracy, linearity and sensitivity of the developed HPLC method. In chapter three, we focused on the HPLC method from the EP for the determination of related substances in ropinirole hydrochloride. The pharmacopeial method did not separate two pairs of impurities and the analysis time was too long. A UHPLC analytical technique was used to develop a new method. Due to a relatively large number of CMPs (buffer molarity, buffer pH, ratio of methanol in the mobile phase, column temperature, initial ratio of organic phase and gradient slope) we performed 19 experiments according to a fractional factorial screening design. Using DoE, we determined the effects of factors, among which the ratio of methanol in the mobile phase, column temperature and gradient slope turned out as highly significant. In this step, we already determined the optimal values for the three less significant factors. In the method optimization step, we performed 17 experiments according to a central composite face-centered response-surface design. Using the MLR method, we established a statistically significant mathematical model with very good prediction abilities. Using the Monte Carlo simulations method, we determined two design spaces according to the elution order of impurity C and unknown degradation impurity. For the final method, we chose the optimal combination of factors from the wider design space and confirmed the suitable precision, accuracy, linearity and sensitivity of the developed UHPLC method. Using AQbD, we developed and optimized two new robust methods, which eliminate the issues of the existing pharmacopeial methods. Using DoE reduced the number of experiments that would otherwise be performed if using OFAT approach, and that the established mathematical model allowed us to obtain detailed information on the effects of factors and factor interactions on a particular response, which is impossible to determine using OFAT. Using DoE, we determined the design space. In the case of ropinirole hydrochloride, we significantly shortened the analysis time and reduced organic solvent consumption, which is in line with current trends of analytical chemistry. For both cases, we have proven the effectiveness of the AQbD approach to the development and optimization of HPLC methods.
|
