Fast label-free live imaging with FlowVision reveals key principles of cancer cell arrest on endothelial monolayers

The rapid, transient, and unpredictable nature of interactions between circulating cells and the endothelium challenges the investigation of these events under flow conditions. Here, we developed an imaging and image-analysis framework called FlowVision, which integrates fast, bright-field live-cell imaging with deep-learning-based image analysis to quantitatively track cell landing and arrest on an endothelial monolayer under physiological flow conditions. Using FlowVision, we find that pancreatic ductal adenocarcinoma (PDAC) cells exhibit variable adhesion strength and flow sensitivity. Remarkably, some PDAC cells demonstrate comparable endothelial engagement to leukocytes, preferentially arresting at endothelial junctions, providing them access to the underlying basal extracellular matrix. PDAC cells attach and form clusters in areas with high expression of the endothelial CD44 receptor. Targeting CD44 using siRNA, function-blocking antibodies, or degrading its ligand, hyaluronic acid (HA), strongly reduces PDAC cell attachment. Overall, our label-free live-imaging approach demonstrates that cancer and immune cells share both common and unique features in endothelial adhesion under flow, and allows identification of CD44 and HA as key mediators of PDAC cell arrest.