@phdthesis{Walther2023,
  author    = {Walther, Rasmus},
  title     = {Analysis of weakly chromophore impurities by means of liquid chromatography coupled with charged aerosol detection and mass spectrometry},
  doi       = {10.25972/OPUS-32186},
  url       = {http://nbn-resolving.de/urn:nbn:de:bvb:20-opus-321862},
  school      = {Universit{\"a}t W{\"u}rzburg},
  year      = {2023},
  abstract  = {In all the projects presented, it is evident that the selection of suitable separation conditions is only one side of the coin. Equally crucial in the development of methods for the quality assessment of APIs/drugs is the right detection system. The application of CAD as an alternative to UV detection at low wavelength of the two weak chromophore main degradation products of the very polar, zwitterionic API carbocisteine requires the volatility of the mobile phase. Therefore, as a substitute for the non-volatile ion pairing reagent tetrabutylammonium hydroxide (TBAOH), six different volatile alkylamines as well as a RP/SAX mixed-mode column were evaluated. The best selectivity and separation performance comparable to TBAOH was achieved with the RP/SAX column and a mixture of formic acid and trifluoroacetic acid. For the simultaneous optimisation of the evaporation temperature of the CAD as a function of two chromatographic parameters, a central composite design was chosen and the "desirability function" was subsequently applied for modelling. In addition, column bleeding was investigated with a second RP/SAX column (different batch) with the result that the acetonitrile percentage had to be adjusted and preconditioning by injection of concentrated samples is essential. The final mixed-mode method was finally validated with both columns according to the ICH Q2 (R1) guideline. Based on this, an MS-compatible method was developed with little effort using an identical RP/SAX column in UPLC dimension for the untargeted analysis by HRMS of two carbocisteine-containing prototype syrup formulations. For a comprehensive characterisation, HRMS and MS/HRMS data were recorded simultaneously by information dependent acquisition mode. Based on the exact masses, isotope patterns and an in silico plausibility check of the fragment spectra, the prediction of the structures of the unknown impurities was possible. In both syrup samples, which had been stored for nine months at 40 °C and 75 \% r.h., two additional impurities of carbocisteine (i.e. lactam of the sulfoxides and disulphide between cysteine and thioglycolic acid) were identified by comparison with the corresponding prototype placebo samples using general unknown comparative screening. In addition, the formation of Maillard products by binary mixtures with 13C-labelled sugars was revealed in the sucrose-containing formulation. For the promising hyphenation of the UV detector with the CAD for the simultaneous detection of all UV-active impurities of the cholesterol-lowering drug simvastatin and the only weak chromophore dihydrosimvastatin, the Ph. Eur. method had to be adapted. Besides replacing phosphoric acid with trifluoroacetic acid, the gradient also had to be adjusted and a third critical peak pair was observed. Based on validation experiments (according to the ICH Q2 (R1) guideline), the suitability of the CAD for sensitive detection (LOQ = 0.0175 \% m/m) was proven.  To further investigate the robustness of the adapted method and CAD, a Plackett-Burman design was chosen. None of the factors had a statistically significant effect on the S/N of the CAD in the ranges tested. Regarding the three critical peak pairs, on the other hand, the factors to be controlled were statistically established, so that a targeted correction is possible if the system suitability test is not passed. The idea of employing a hyphenated UV-CAD system was finally applied to the structurally closely related lovastatin and its specified impurity dihydrolovastatin. Here, the CAD showed a significantly better S/N compared to the compendial UV detection at 200 nm. The suitability of CAD for the analysis of non-volatile fatty acids in polysorbate 80 (PS80) as favourable alternative to the Ph. Eur. GC method (no time-consuming, error-prone and toxic derivatisation) has already been demonstrated. The aim of this project was therefore to develop a robust method with a focus on the AQbD principles, which can be used for the analysis of other excipients with similar fatty acid composition. After the definition of the analytical target profile and a risk assessment by means of an Ishikawa diagram, a suitable C18 column and the chromatographic framework conditions (formic acid concentration and initial/final gradient conditions) were selected after only few preliminary runs. The remaining critical method parameters were then investigated with the help of DoE and RSM. Using the obtained model equations, Monte Carlo simulations were performed to create the method operable design region as a region of theoretical robustness. After validation according to ICH Q2 (R1), the fatty acid composition of a magnesium stearate batch was successfully analysed as a further application example in addition to PS80. The CAD was able to prove its potential in all the issues investigated in the context of this doctoral thesis. As a cost-effective alternative compared to MS instruments, it thus closes a gap in the quality assessment of APIs or excipients without a suitable chromophore. The easy method transfer to (HR)MS instruments also allows for a unique degree of sample characterisation through untargeted approaches in case of new impurities. For resource- and time-efficient work, the possibilities and limitations of software tools for method development and data evaluation as well as the application of risk-based approaches such as AQbD should also be considered.},
  subject      = {Carbocistein},
  language  = {en}
}