At first mostly dedicated to molecular analysis, microfluidic systems are rapidly expanding their range of applications towards cell biology, thanks to their ability to control the mechanical, biological and fluidic environment at the scale of the cells. A number of new concepts based on microfluidics were indeed proposed in the last ten years for cell sorting. For many of these concepts, progress remains to be done regarding automation, standardization, or throughput, but it is now clear that microfluidics will have a major contribution to the field, from fundamental research to point-of-care diagnosis. We present here an overview of cells sorting in microfluidics, with an emphasis on circulating tumor cells. Sorting principles are classified in two main categories, methods based on physical properties of the cells, such as size, deformability, electric or optical properties, and methods based on biomolecular properties, notably specific surface antigens. We document potential applications, discuss the main advantages and limitations of different approaches, and tentatively outline the main remaining challenges in this fast evolving field.

We propose a strategy for optimizing distribution of flow in a microfluidic chamber for microreactor, lateral flow assay and immunocapture applications. It is aimed at maximizing flow throughput, while keeping footprint, cell thickness, and shear stress in the distribution channels at a minimum, and offering a uniform flow field along the whole analysis chamber. In order to minimize footprint, the traditional tree-like or "rhombus" design, in which distribution microchannels undergo a series of splittings into two subchannels with equal lengths and widths, was replaced by a design in which subchannel lengths are unequal, and widths are analytically adapted within the Hele-Shaw approximation, in order to keep the flow resistance uniform along all flow paths. The design was validated by hydrodynamic flow simulation using COMSOL finite element software. Simulations show that, if the channel is too narrow, the Hele-Shaw approximation loses accuracy, and the flow velocity in the chamber can fluctuate by up to 20%. We thus used COMSOL simulation to fine-tune the channel parameters, and obtained a fluctuation of flow velocity across the whole chamber below 10%. The design was then implemented into a PDMS device, and flow profiles were measured experimentally using particle tracking. Finally, we show that this system can be applied to cell sorting in self-assembling magnetic arrays, increasing flow throughput by a factor 100 as compared to earlier reported designs.

A new generation of the Ephesia cell capture technology optimized for CTC capture and genetic analysis is presented, characterized in depth and compared with the CellSearch system as a reference. This technology uses magnetic particles bearing tumour-cell specific EpCAM antibodies, self-assembled in a regular array in a microfluidic flow cell. 48,000 high aspect-ratio columns are generated using a magnetic field in a high throughput (>3 ml h(-1)) device and act as sieves to specifically capture the cells of interest through antibody-antigen interactions. Using this device optimized for CTC capture and analysis, we demonstrated the capture of epithelial cells with capture efficiency above 90% for concentrations as low as a few cells per ml. We showed the high specificity of capture with only 0.26% of non-epithelial cells captured for concentrations above 10 million cells per ml. We investigated the capture behavior of cells in the device, and correlated the cell attachment rate with the EpCAM expression on the cell membranes for six different cell lines. We developed and characterized a two-step blood processing method to allow for rapid processing of 10 ml blood tubes in less than 4 hours, and showed a capture rate of 70% for as low as 25 cells spiked in 10 ml blood tubes, with less than 100 contaminating hematopoietic cells. Using this device and procedure, we validated our system on patient samples using an automated cell immunostaining procedure and a semi-automated cell counting method. Our device captured CTCs in 75% of metastatic prostate cancer patients and 80% of metastatic breast cancer patients, and showed similar or better results than the CellSearch device in 10 out of 13 samples. Finally, we demonstrated the possibility of detecting cancer-related PIK3CA gene mutation in 20 cells captured in the chip with a good correlation between the cell count and the quantitation value Cq of the post-capture qPCR.