Usefulness of D-optimal designs and multicriteria optimization in laborious analytical procedures. Application to the extraction of quinolones from eggs.

An analytical method has been developed to extract ciprofloxacin and enrofloxacin from eggs. The aim of this work is to determine the experimental conditions of extraction providing high recoveries with small standard deviations. An experimental design based on the D-optimality criterion and replicated three times was built to evaluate the effect of five factors related to the extraction which is the most inaccurate stage of the procedure. This non-classical design is needed because there are several practical constraints: (i) the extraction procedure is time-consuming, quinolones are not stable and the design must be performed in a single working session. (ii) The tube capacity of the centrifuge is 6, so the number of experiments will be 6 or a multiple of 6. In the optimal experimental conditions, the extraction is performed once with 5 ml of methanol. Then, fatty acids are removed with a mixture of hexane/ether. Analytes are finally separated and detected by HPLC-fluorescence without the additional step of purification by solid-phase extraction (SPE). Under these conditions, the mean recovery is 64% and 70% and the standard deviation 5% and 4% for ciprofloxacin and enrofloxacin, respectively. The capability of decision, CCalpha, is 3.1 and 2.8 microg kg(-1) of ciprofloxacin and enrofloxacin, respectively. The capability of detection, CCbeta, is 7.8 and 7.0 microg kg(-1) of ciprofloxacin and enrofloxacin, respectively. In both cases the probabilities of false positive, alpha, and of false negative, beta, were fixed at 0.05.

Robustness of the extraction step when parallel factor analysis (PARAFAC) is used to quantify sulfonamides in kidney by high performance liquid chromatography-diode array detection (HPLC-DAD).

The robustness of a multiresidue method has been analysed for the extraction and quantification of sulfamethoxypyridazine, sulfamethoxazole and sulfadimethoxine in porcine kidney by HPLC-DAD through a Plackett-Burman design. Two experimental responses were examined, the mean recovery from three replicates (accuracy) and their standard deviation (precision). Three factors were tested: the volume of phosphoric acid (pH) added in the extraction step, the time used for passing the sample through the solid-phase extraction cartridge (flow rate) and methanol volume to elute the analytes from the cartridge. Due to the non-specificity of the chromatograms (unknown matrix interferences coelute with each sulfonamide) the PARAFAC model was employed to evaluate the concentration recovered in the experiments of the Plackett-Burman design as well as to identify the spectra of the substances according to the criteria set in the European Decision 2002/657/EC for the analysis of residues. The extraction step was concluded to be robust to the recovery and the standard deviation of all three analytes.