Hepatocellular carcinoma (HCC) causes ~ 600,000 deaths annually, making it the second most deadly cancer form. HCC is classified into five stages and for the intermediate HCC treatment, the two most commonly used drug delivery systems (DDSs) are lipiodol-based emulsions and drug-eluting beads. The aims of this thesis were to develop in vitro methods suitable for studying these DDSs. It is important to investigate the release mechanisms and release rates with relevant in vitro methods, as this can improve the understanding of the in vivo performance. Miniaturized in vitro methods with sample reservoirs separated from the release medium by a diffusion barrier were developed and shown to be suitable for studying drug release from particle DDSs (Paper I). In Paper II these methods were further developed and used to study the release of doxorubicin (DOX) from the clinically used drug-eluting beads. DOX release rates were affected by the method set-up and the characteristics of the release medium. The choice of method and volume of release medium could improve the in vivo-likeness of the in vitro release profiles. Applied theoretical models suggested a film-controlled type of DOX release mechanism from the beads when self-aggregation, DOX-bead interaction, and DOX deprotonation were taken into account.

A micropipette-assisted microscopy method was used to further improve the understanding of the release mechanism of amphiphilic molecules from the beads (Paper III). A detailed analysis suggested an internal depletion-layer model dependent on molecular self-aggregation for the release. It was further suggested that a simple ion-exchange mechanism is unrealistic in physiological conditions.

The important pharmaceutical factors for the emulsion-based formulations were investigated in Paper IV. DOX solubility, lipid phase distribution, and emulsion stability increased when the contrast agent iohexol was added. Also, an increase in release half-life (h) was observed from emulsions with iohexol.

The in vitro methods and theoretical models presented in this thesis can be used during development and optimization of future DDSs.