Optimized preclinical models for human pancreatic adenocarcinoma therapy research

Abstract

15019 Background: Efforts to find new therapies for human pancreatic ductal adenocarcinoma (PDAC) have not resulted in clear improvements on patient survival. Better knowledge of resistance mechanisms and redefiniton of molecular targets is essential to design more efficient therapies. The multifactorial origin of PDAC points to combined strategies as the therapy of choice, though the effective development of such strategies is hampered by the lack of optimal preclinical models. We have generated and validated optimized human PDAC models by direct implantation of fresh tumoral tissue into the pancreas of athymic mice. Methods: Thirteen pancreatic adenocarcinoma specimens from PDAC patients were obtained by surgical resection. From each specimen, several 10 mg-fragments were used to generate the corresponding intrapancreatic xenografted tumours. Eleven human PDAC orthotopic models have been successfully generated and perpetuated by succesive passages (up to 4). Histological and molecular analyses of both primary and xenografted tumors have been performed by tissue- array, western-blot and DNA sequentiation. Results: Initial engraftment rate ranged from 20 to 100% (mean 59%) and it improved with succesive passages (mean 76% at second and 90% at third generation). Ki67 expression and degree of differentiation in primary tumors correlated with xenograft growth kinetics. Furthermore, their spontaneous metastatic behaviour fairly reproduced the original patient dissemination patterns. Xenografted tumors kept the original architecture and expression patterns of common PDAC markers. Efficacy of several agents was tested on different xenografted tumors, validating this model and underlining its utility to define future therapeutic strategies for drug development and clinical trials. Conclusions: The orthotopic models described here are, probably, the closest resemblance to a patient clinical setting since they preserve human pancreatic structures, genotypic features and biological behaviour. From their use, biological relevant data could be drawn for future clinical trials and for testing new agents and new drug combinations since they represent, very likely, the most reliable animal models at present. No significant financial relationships to disclose.