Predictive value of genetically engineered endogenous mouse models in preclinical therapeutic studies

Pancreatic ductal adenocarcinoma (PDAC) remains one of the most lethal malignancies with an alarming resistance to both chemotherapeutics and targeted therapy approaches. Genetically engineered mouse models (GEMM) of PDAC reflect molecularly and pathophysiologically the carcinogenesis and cancer progression observed in humans, thus being excellent tools for preclinical evaluation of new therapies.

To successfully predict and characterize the effect of chemotherapies and targeted approaches in PDAC, we use a complex and highly aggressive GEMM of endogenous PDAC, Ptf1a +/Cre Kras +/LSL-G12D p53Lox/Lox (CKP) mice. To do so, we established a highly usable GEMM-based therapy platform including multi-parametric MR imaging using CKP mice, in which tumors develop within the first 6 weeks of age (Bardeesy et al, 2006).

Animals receive a T2w scan on a 3T clinical MRI scanner for detection and staging of solid tumors for study enrolment at defined inclusion criteria (tumour volume 200 – 400 mm3). Mice get a weekly scan throughout the study until endpoint criteria are met. Upon reaching the endpoint criteria, mice are sacrificed and tumor material is processed for histopathological and RNA/Protein analysis, isolation and culturing of primary tumor cells. Tumors are assessed macro- and microscopically and graded. Proliferation, apoptosis and analysis of respective downstream effectors are characterized. Further comprehensive analysis depending on the respective phenotypical features is then carried out. Primary cell lines are generated for further functional and molecular characterization including but not limited to cell viability, drug sensitivity, expression and methylation profiling.

Through this platform, we have successfully been able to target key signaling pathways and to identify promising novel chemotherapeutic and targeted combinations for PDAC. Here we want to highlight the vital need to develop improved preclinical tools to characterize individual tumors throughout the course of treatment. We propose that such a platform is highly relevant and usable for predicting responses to chemotherapeutic and targeted agents.