Development and validation of a microfluidic platform for in vitro studies on pancreatic cancer and drug response

Microfluidic platforms offer a means of combining controlled chemical gradients with three-dimensional tumor models for in vitro drug testing. This thesis presents the design, numerical optimization, and functional validation of an innovative multi-layer microfluidic platform for studying the response of PANC-1 pancreatic cancer spheroids to chemotherapeutic agents. The device integrates a concentration gradient generator, an array of spheroid-hosting microwells and an outlet collection system, enabling simultaneous exposure to six nominal drug concentrations. Device fabrication by photolithography and resin 3D printing was evaluated through profilometric analysis, with photolithographic masters providing more uniform channel heights. Numerical simulations predicted the formation of the intended concentration gradient and solute transport into GelMA-filled microwells. Experimental tests with colored tracers, FITC-dextran and Trypan Blue supported gradient generation, with quantitative outlet measurements approaching the nominal concentration profile. To better replicate the dense extracellular matrix characteristic of pancreatic ductal adenocarcinoma, the platform architecture is engineered to integrate gelatin methacryloyl (GelMA) hydrogels within a microwells layer hosting 3D PANC-1 tumor spheroids. ATP-based assays showed concentration-dependent reductions in viability following doxorubicin and 5-fluorouracil treatment. Exploratory comparisons with matched static cultures showed higher viability under perfusion at the same nominal doses, highlighting the need to distinguish the concentrations generated by the network from the effective exposure of spheroids within the microwells. Preliminary experiments also demonstrated the feasibility of embedding spheroids in GelMA within the device and maintaining viable cells over 48 hours under static conditions; drug testing in this configuration remains to be investigated. Overall, this work establishes the feasibility of integrating concentration gradient generation with on-chip spheroid culture and drug-response measurements. Further characterization of local drug exposure and experimental variability is needed to support quantitative interpretation and extension to matrix-embedded tumor models.