High-throughput, imaging based mechanical phenotyping of prostate cancer cells

Summary form only given. Prostate cancer (PCa) is the second most common type of cancer in the world and the fifth highest cause of cancer-related deaths in men [1], with the highest prevalence in the United States and Western Europe. The current diagnostic gold standards are controversial and typically lead to over-diagnosis [2]. Blood tests are commonly used to check the level of the prostate specific antigen (PSA), when in fact this antigen is organ-specific but not cancer-specific. Lack of clarity over diagnosis as well as prognosis leads to large numbers of unnecessary treatments, which are highly invasive and with a range of unpleasant side effects. New methods are therefore required to improve the existing clinical outcomes at both the diagnostic and the prognostic level.Here we present a hydrodynamic stretcher [3], with time-resolved capabilities, in which single cells are deformed hydrodynamically in a microfluidic device. We report a dynamic mechanical phenotyping analysis of conventional prostate cell lines: DU145, characterised by moderate metastatic potential [4] and PNT2, commonly used as healthy controls. We focus on their mechanical characterisation using a novel microfluidic hydrodynamic stretching device. Each cell is imaged using high-speed microscopy (300,000 frames per second) during its interaction with a pinching flow over successive frames. An ad hoc automatic tracking algorithm enables us to quantify and record the cellular Roundness [5] temporal profiles, that is a measure of how each cell changes its shape over time as a response to the applied stress. These profiles are then used as a biomarker to identify difference between the two samples. Finally, we classify the two cell lines based on their time-resolved Roundness using machine learning approaches.