The Chemometric Study and Quantitative Structure Retention Relationship Modeling of Liquid Chromatography Separation of Ziprasidone Components

Abstract

Optimization of the experimental conditions of a novel HPLC method for determination of the impurity levels with ziprasidone (in bulk substance and pharmaceutical dosage forms) was performed with use of Multi-Layer Perceptron (MLP) Artificial Neural Networks (ANN) and Response Surface Plots. The obtained experimental conditions were further used to test a set of 20 reversed-phase columns for their selectivity towards ziprasidone components by use of the principal component analysis (PCA) and hierarchical clustering analysis (HCA). The obtained HPLC retention times of ziprasidone and its impurities (Imp I-V) along with the computed molecular parameters of the examined compounds were further used in the Quantitative Structure Retention Relationship (QSRR) study. The performed QSRR study has selected the LogD(pH) (1.5), LogD(pH 2.5), LogD(pH 4.0), LogP, MS, and SAS parameters as descriptors of the chromatographic behavior of ziprasidone components. The developed QSRR model can be very useful in the t(R) prediction for the ziprasidone derivatives (impurities, degradation products, and metabolites). As the performed LC-MS study of the test solution has confirmed that the unknown impurity (t(R): 11.270 min) in the test solution is the TS1, one from two candidates predicted by QSRR (TS1 and TS5), the high prediction potential of the created QSRR models has been proved.