Objective

Drug screening aimed at personalised therapy using 3D tumour models is challenging due to the limited number of cells contained in most biopsies. Microfluidic technologies, enabling greater control of fluid behaviour and cell distribution, can provide medium- to high-throughput solutions to facilitate physiologically relevant drug screening in a cost-efficient manner. We developed a microfluidic platform to miniaturise screening of spheroids, that finds application in a variety of drug testing scenarios. In particular, the platform provides unprecedented capability to maximise drug testing when using tissue derived from cancer patient biopsies. Uniquely, the platform generates long-lasting, stable drug concentration gradients across arrays of hundreds of spheroids without needing external instrumentation for fluid actuation. A typical drug screening assay could generate up to 22 drug concentration-response curves from a single biopsy, utilising readouts of spheroid growth, viability and drug efficacy. As proof of concept screening for personalised therapy, prostate biopsy tissue was grown as a heterogeneous co-culture and consequently used for the generation of thousands of spheroids in the microfluidic system. Spheroids were cultured for 3 to 5 days prior to exposure to a panel of commonly used drugs for prostate cancer. A network of microfluidic channels generated a stable drug concentration gradient for at least 16 hours across arrays of 240 spheroids. In house developed software was used to process brightfield and epifluorescence microscopy images to obtain readouts identifying spheroid size, shape and viability, creating either 5 or 8 point concentration response curves from at least 24 spheroids per concentration. Comparison between prostate cancer cell lines and biopsies from 2 different prostate cancer patients showed that biopsy-derived spheroids were more resistant to treatment than prostate cell lines. Spheroids from both patients were sensitive to docetaxel, but resistant to enzalutamide, despite the presence of intact androgen receptors in both biopsies. These results demonstrate that this technology has great potential to impact assessment of patient-specific drug resistance, using a cost-effective approach based on 3D tumour models.