Comparison of Various Multivariate Mathematical-Statistical Methods for the Evaluation of the Retention Behaviour of β-Cyclodextrin Polymer Coated Silica Column

Abstract

The retention of 17 monoamine oxidase inhibitory drugs was determined on a β-cyclodextrin polymer (βCDP)-coated silica column using acetonitrile - 0.05 M K₂HPO₄ eluent. The relationship between retention capacity and the physicochemical parameters of solutes determined with various adsorptive and reversed-phase TLC methods was evaluated by principal component analysis followed by two-dimensional non-linear mapping, vari-max rotation around two axes and cluster analysis. The HPLC and TLC retention parameters of the drugs were widely separated on the two dimensional nonlinear map of principal component loadings indicating the unique retention characteristics of βCDP column. Significant linear correlations were found between the rotated principal component loadings and the coordinates of their two-dimensional non-linear map. The better separation of variables on the two dimensional maps suggested that the principal analysis followed by two-dimensional non-linear mapping or by varimax rotation is a more useful method for evaluation of large data matrices in chromatography than the one dimensional cluster analysis, although the latter uses markedly less computer time.

Key Words:

• β-cyclodextrin coated silica column
• principal component analysis
• cluster analysis
• Varimax rotation