Application of a column classification method in a selectivity study involving caffeine and its related impurities Jaros"aw Szulfer a , Alina Plenis b , Tomasz Ba˛czek b,n a Polpharma SA Pharmaceutical Works, Pelplin ´ska 19, 83-200 Starogard Gdan ´ski, Poland b Department of Pharmaceutical Chemistry, Medical University of Gdan ´sk, Hallera 107, 80-416 Gdan ´sk, Poland article info Article history: Received 22 February 2012 Received in revised form 4 June 2012 Accepted 6 June 2012 Available online 18 June 2012 Keywords: Column classification system Factor analysis LC separations Caffeine Impurities abstract In this paper a comparative study of RP-LC column selectivity as obtained by the classification method of the Katholieke Universiteit Leuven (KUL method) and the selectivity obtained in real pharmaceutical analysis is reported. The separation of caffeine and its respective impurities was performed on 35 brands of stationary phases in accordance with the method prescribed in the European Pharmacopoeia (Ph. Eur.). Evaluation concerned the probability of appropriate column selection related to the selection of two different stationary phases for reference. The comparison was based on the traditional correlation of the F-values with the results of a system suitability test (SST) for the columns, as well as an application of a factor analysis (FA) for graphical visualisation of the differences and similarities between the stationary phases established against four test chromatographic parameters provided by the KUL method and the retention parameters of the compounds of interest describing the column performance test. The obtained results confirmed that the class of the stationary phases selected according to the chromatographic test parameters gave comparable separation for caffeine and its impurities. & 2012 Elsevier B.V. All rights reserved. 1. Introduction Reversed-phase liquid chromatography (RP-LC) today prob- ably belongs to the separation techniques most frequently used in pharmaceutical and biomedical analyses [1,2]. However, the hundreds of commercially available columns offered in the market in combination with the great variety of possible chro- matographic systems have turned the selection of the stationary phase suitable for real application into a challenging task. The choice is additionally complicated by the fact that the RP-LC phases often belong to the same chemical class, which may suggest their similar chromatographic properties. In practice, the polar and ionic properties of the RP-LC phases responsible for the secondary intermolecular interaction mechanisms cause the stationary phases to give RP-LC columns a unique character. Therefore, the differences between the physical and/or chemical properties of stationary phases cause the analysts to confront the problem of column selection for a given separation. The same difficulties can be encountered by analysts who perform separa- tions in accordance with the official monographs of the European Pharmacopoeia (Ph. Eur.) [3] and the United States Pharmacopeia (USP) [4]. The monographs report numerous chromatographic methods, mainly under RP conditions, but provide merely general descriptions of the stationary phases to characterise the suitable column, including e.g. the chain length, end-capping, base-deac- tivation, particle size, pore size, and specific surface. More detailed information about the stationary phase can only be found for the recently developed Ph. Eur. and USP monographs on their websites [5,6]. It may just as well happen that the name of the required column will be known but its application for a given analysis might be impossible because the prescribed sta- tionary phase is not available in the laboratory or simply not attainable from the market any longer. Sometimes, the chromato- graphic behaviour of the RP stationary phases can alter because of the storage time or usage causing their potentially different selectivity [7]. In such a case it would be helpful to have a replacement of a suitable alternative offering separation ‘‘equiva- lent’’ to the original column. Therefore, many analysts expect reliable test methods should characterise the RP-LC columns so as to solve the problem of non-suitable column selection. For this reason, extensive investigations involving several chromato- graphic column tests were conducted over the last two decades [8–14]. The interesting approaches reported in the published papers were as follows: the mathematical models including the hydrophobic-subtraction (HS) model proposed by the Snyder and Dolan group [15–17], the linear solvation–energy relationship (LSER) delivered by Abraham and Sa ´ ndi and Szepesy [18,19], the quantitative structure–retention relationships (QSRRs) provided Contents lists available at SciVerse ScienceDirect journal homepage: www.elsevier.com/locate/talanta Talanta 0039-9140/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.talanta.2012.06.015 n Corresponding author. Tel.: þ48 58 3493133. E-mail address: tbaczek@gumed.edu.pl (T. Ba˛czek). Talanta 99 (2012) 492–501