Feasibility of implantable microdosing devices to perform in vivo pharmacotyping in pancreatic cancer.

Abstract

742 Background: Current biomarkers offer modest predictive capacity for drug sensitivity testing in pancreatic cancer. Patient-specific pharmacotyping offers the potential for a more direct, personalized prediction model. Recently, an implantable microdevice was developed to facilitate in situ multi-agent pharmacotyping and has since entered clinical trials in multiple cancer types, but had not yet been tested in pancreatic cancer. Here we evaluated the use of a microdosing device (microdevice) in a mouse xenotransplant model of pancreatic cancer as proof of principle to inform future translation to the clinic. Methods: A microdevice consists of a cylindrical frame and 18 wells embedded with standard chemotherapy agents (5 FU, gemcitabine, SN 38, paclitaxel, oxaliplatin), a non-toxic polyethylene glycol (PEG) matrix as control, and doxorubicin, the auto-fluorescent properties of which allow us to determine the distance of drug diffusion for each tumor. Devices were implanted into murine xenotransplanted human organoids (hM1A and hM1E Table 1). Microdevices retrieved at 4h, 8h, and 24h post-implantation together with tumors were fixed in 10% formalin. The tissue was embedded in paraffin and sectioned parallel to the device. A distance-dependent drug diffusion rate and concentration profiles with doxorubicin were evaluated. Co-immunohistochemistry (IHC) for cleaved caspase-3 (CC3), gamma H2AX (apoptosis markers), and human mitochondria (human cancer cell marker) was performed in the area of drug diffusion to determine the cytotoxic effects of chemotherapy. Results: 15 animals were enrolled in the experiment with 100% postoperative survival. One device per animal was placed, and all devices were successfully retrieved. Microdevice insertion and exposure to PEG had no effect on tumor integrity. At 250-350µm from the well, we observed intensity from doxorubicin, suggesting drug diffusion. The signal on differential response to drugs was detected as early as 4 hours. Conclusions: Utilization of implantable microdevice to predictcancer drug sensitivity is feasible in organoid transplantation animal models of pancreatic cancer. Future work is needed to expand this model to other cancer drugs and biomarkers to evaluate the cancer drug sensitivity and to test the feasibility of this model to study the effects of chemotherapy on the tumor microenvironment. [Table: see text]