Analysis of the retention of tetracyclines on reversed-phase columns: Chemometrics, design of experiments and quantitative structure-property relationship (QSPR) study for interpretation and optimization.

In this study, design of experiments was applied for the analysis of 6 reversed phase U-HPLC columns used for the separation of four tetracyclines (TCs): tetracycline, doxycycline, chlortetracycline and oxytetracycline in different elution conditions. In a first part, a fractional factorial design (2) was used to study the influence of four chromatographic parameters: column temperature, pH, flow rate and composition of the mobile phase (i.e. nature of the solvent used as the organic modifier), on the quality of the separation, which was evaluated in terms of peak width and resolution between two pairs of TCs. This experimental design revealed that the nature of the solvent: acetonitrile (ACN) or methanol (MeOH), and the mobile phase flow rate were the two main factors actually having the most influence on the quality of the separation. Moreover, these two factors presented an antagonistic influence according to the response considered: peak width or peak resolution. In order to understand this behavior, a Doehlert design was performed in the second part. It consisted in modeling the evolution of responses as a function of the two main factors: nature of the composition of the mobile phase (mix of ACN and MeOH, from 100% ACN to 100% MeOH) and mobile phase flow rate (from 0.3 to 0.8 mL min). For all the reversed phase columns studied, an inversion of the elution order of TCs and an increase of the retention factors was observed according to the composition of the organic mixture at the end of the gradient. To understand the modification of the interactions implied in the various retention modes related to the selectivity of the organic solvents used, a quantitative structure-property relationship (QSPR) study was achieved. In this final study, the molecular descriptors of each TCs were connected to its retention factor.