Cancer is a very heterogeneous disease that effects millions worldwide. A common challenge in drug efficacy trials is that widely used cancer models largely lack many attributes that are needed to mimic in-vivo behaviour, which results in less relevant results and slows the progression of drugs to clinical trials. These traditional models commonly consist of two dimensional (2D) cultures in wells, flasks or plates, with no environmental control and a lack of mechanical and chemical stimuli from other cells. Consequently, it is widely accepted that the 2D models behave differently than three dimensional (3D) in-vitro tumour models, which in turn behave differently compared to cancer tumours in-vivo. In the context of this thesis, the work described concentrates on improving the 2D cancer model by initiating drug resistant regions by incorporating dissolved oxygen (DO) control and measurement.

One of the main known environmental differences between the tumour environment in-vivo and the lab models is the DO environment. Low oxygen, or hypoxia, occurs in cells in-vivo when oxygen usage exceeds influx. This imbalance can be such that the concentration within the tissue becomes hypoxic over time and can contain as little as 5 % of the oxygen usually present in healthy tissue counterparts. Hypoxic regions are common in cancerous tissue in-vivo due to abnormally high cell-proliferation rates that increases oxygen use where damaged or absent vasculature in the tumour core decrease its supply. This leads to DO heterogeneity over hundreds of microns, a feature which is not captured in common 2D cancer models. Incorporation of spatial DO control into the lab model would be a significant step towards a more complete cancer model that behaves in a more life-like manner, as hypoxia has a profound effect on the behaviour of cells and on the efficacy of drugs. In general, hypoxia in-vivo is associated with aggressive phenotypes, recurrence of disease, shorter lifespan before recurrence and drug resistance. Therefore, it is crucial that oxygen be included as a controlled parameter in 2D models of cancer during drug discovery and screening, so that the most relevant results can be gathered before transition to the clinic.

In this work, DO concentrations were controlled in the cell-culture media in which cells were cultured by a passive diffusion-driven process in an off-chip gas exchanger. Oxygen measurement was facilitated by flow-through optical oxygen sensors and a flu- orescent polystyrene (PS)-based fluorescent oxygen sensor substrate, with incorporated oxygen-sensitive fluorophore platinum(II) octaethylporphyrinketone (PtOEPK). This 550 nm thick PS/PtOEPK thin-film provided a cell-culture substrate with non-invasive oxygen sensing capabilities. The microfluidic system was capable of exposing a 2D cul- ture of cells of any type to a cross-stream DO gradient ranging from hypoxia (<4 mg/L) to hyperoxia (>40 mg/L) for a maximum of three hours. The response of the cells to the gradient was characterised with visible light microscopy, several types of cell via- bility staining and immunostaining of the hypoxia activated factor hypoxia inducible factor 1 (HIF-1)α. Culturing two cell lines on the PS/PtOEPK sensor film was found to decrease cell viability in a cell-line dependant manner. Ishikawa cells were found to have a lower viability than SKOV-3 after exposure to shear forces from perfusion in a microchannel and SKOV-3 cells were found to decrease in viability more than Ishikawa when exposed to controlled DO conditions. Additional to designing a chemical-free DO measurement and control system, a novel method was developed to increase can- cer cell adherence to the oxygen sensor substrate by air plasma exposure followed by polyvinylpyrrolidone (PVP) treatment. This allowed the PS/PtOEPK film to be used as a cell-culture substrate in-chip with continuous perfusion, and has never before been used in this way.

The work presented in this thesis provides a step towards a 2D microfluidic cancer model incorporating DO control and measurement towards adding in-vivo-like complex- ity to the traditional 2D lab models. The presented platform provides a drug-screening tool for the culture of cancer cells in an in-vivo-like microenvironment.