Contributions to faster development and validation of pharmaceutical analytical separation methods : experimental design in method optimization and method validation

When developing fast methods, two approaches are possible : the optimization of classical separations and the evaluation of incomplete separations. The subject of this project is to study fast but classical separations. Moreover, research will be done to develop techniques or strategies to increase the speed of method development, parts of method validation and the assay procedure, without requiring the application of chemometric techniques such as deconvolution and warping. Moreover, the use of minimal and/or less frequently used experimental designs in the development and validation/robustness testing of fast separation methods will be evaluated.
The use of experimental design in method validation (mainly robustness testing) will be examined. The use of the variance of a response of supersaturated designs as a measure for method robustness will be studied, as well as the estimation of factor effects in these designs. Since minimal screening designs (Fractional Factorial or Plackett-Burman designs) do not always have enough interaction or dummy effects for a statistical interpretation based on those negligible effects, the use of randomization tests as statistical interpretation method to identify significant effects in experimental designs for robustness testing will be examined. Experimental designs for robustness testing with a different number of experiments will be evaluated and compared : a Plackett-Burman design with 12 or 8 experiments and a supersaturated design with 6 experiments. It will be evaluated whether the same factor effects are found to be significant and whether there is a loss of information when using designs with fewer experiments.
The use of experimental design in method development and optimization will also be evaluated. The use of Process Sigma, Gage R&R studies and an experimental design approach to improve the method capability of an HPLC method for a drug substance assay was evaluated. It was studied which factor had the largest effects on the capability of chromatographic HPLC methods in order to improve their precision and Precision-to-Tolerance ratio. The importance of three-factor interactions and their influence on the fractionation of designs will be evaluated. The estimation of factor effects in uniform designs, in D-optimal designs and Rechtschaffner designs will be examined.
The determination of 4-aminophenol (4-AP) in paracetamol tablets with a blue coating by means of an FIA method was evaluated with 2-channel spectroscopy, PLS and Ridge regression (the last two : only at 2 wavelengths). Also the detection limit and the quantification limit of a fluorimetric method to determine 4-AP are being determined, for comparison with those from the HPLC and FIA methods (first in white paracetamol tablets, later in blue tablets containing a disturbing dye).