Journal of Pharmaceutical and Biomedical Analysis 18 (1998) 57–65 Experimental design for a pharmaceutical formulation: optimisation and robustness 1 B. Campisi a, *, D. Chicco b , D. Vojnovic b , R. Phan-Tan-Luu c a Department of Economics and Commodity Science, Uniersity of Trieste, ia Valerio 6, I -34127 Trieste, Italy b Department of Pharmaceutical Sciences, Uniersity of Trieste, P.le Europa 1, I -34127 Trieste, Italy c LMRE, Centre de St. Je ´ro ˆme, Aenue Escadrille Normandie -Niemen, F -13397 Marseille, Cedex 20, France Received 22 October 1997; received in revised form 29 December 1997; accepted 5 January 1998 Abstract In pharmaceutical industries, the formulator is usually faced with the optimisation of the excipient mixture composition aimed to prepare a product with the required characteristics. Experimental research methodology represents an efficient approach for solving such optimisation problems. Planning mixture experiments using specific designs allows to analyse the blending properties of each mixture component and estimate an empirical model approximating the response of interest as a function of excipient proportions. In this study the evolution of theophylline solubility in a four-component system with constraints was analysed using two mixture design approaches: a classical mixture component proportion approach and a mathematically independent variable ap- proach. An optimal region characterised by high solubility values was found and further explored in order to verify the insensitivity of theophylline solubility to slight variations of the excipient mixture composition. © 1998 Elsevier Science B.V. All rights reserved. Keywords: Experimental mixture design; Component proportion approach; Mathematically independent variable approach; Cosolvent mixture optimisation; Theophylline solubility 1. Introduction A common problem in the preformulation of pharmaceutical dosage forms is the optimisation of the excipient mixture composition aimed to obtain a product with the required characteristics. When the target is to evaluate one or more blend properties, the experimenter can make use of a set of tools and techniques known under the name of experimental research methodology. In general, adopting such an experimental approach means defining the problem which one is going to cope with by determining the objectives, the possible constraints on the compo- nent proportions, and the response variables under study. In this way, the experimental region of interest and the strategy to follow can be defined. A classical mixture problem in pharmaceutical technology is represented by cosolvent systems * Corresponding author. Tel.: +39 40 6767031; fax: +39 40 6763215; e-mail: campisi@uts.univ.trieste.it 1 Presented at the 7th Meeting on Recent Developments in Pharmaceutical Analysis, Island of Elba, Italy, September 16–20, 1997. 0731-7085/98/$ - see front matter © 1998 Elsevier Science B.V. All rights reserved. PII S0731-7085(98)00175-7