Abstract 1668: Mutational landscape and pharmacological profiling of a panel of prostate PDX models including hormone-naïve, hormone-sensitive and castrate-resistant prostate cancer specimens

Prostate Cancer (PCa) is the second most frequent cancer in men worldwide and the fifth leading cause of cancer death with an incidence rate of 13.5%. PCa is driven by multiple genomic alterations, with distinct patterns and clinical implications. These genomic alterations occurring both early and later in the natural history of the disease (ranging from localized disease, initially responsive to androgen deprivation therapy, to Castrate Resistant Prostate Cancers -CRPC) allow classification of PCa in several molecular subtypes with potential clinical relevance. Patient-Derived Xenograft (PDX) models have become the most reliable in vivo human cancer models. Developing such models that capture the biological heterogeneity and mutational landscape of PCa, remains a challenge, but is essential for delivery of precision medicine in metastatic castrate resistant stages. In this study, we present the genomic and transcriptomic landscapes, as well as the pharmacological status of an established bank of seven (7) prostate PDX models ranging from hormone naïve to hormone-resistance PCa specimens. Samples of PCa along with normal corresponding tissues were obtained directly from patients at surgery. Fragments were subcutaneously xenografted into immunocompromised mice to establish PDX models. After the first growth in mice, they were serially passaged in vivo and considered to be established from P3. To ensure model stability, PDX tumors at multiple passages and patients' primary tumors were processed for histological, transcriptomic (Affymetrix U133 plus 2.0 microarray) and STR profile analyses. Genomic characteristics (WES, CNA) were also investigated. Finally, the responses of the PDX models to androgen deprivation and docetaxel were also evaluated. 7 PDX models were successfully established (> P3 in mice) out of 253 primary prostatic tumors collected from surgery. Within those models, one matched pair of responsive adenocarcinoma and neuroendocrine castration-resistant (NE-CRPC) models from the same patient was generated. Histological, transcriptomic and STR profiling validated the stability of the models compared to the parental tumor. The genomic analyses revealed i) the mutational burden rise with the resistance to treatments of the models, correlating with clinical results ii) an increase of metastatic genes loss in the NE-CRPC compared to the corresponding hormone sensitive adenocarcinoma. Furthermore, for all the PDX models generated, genomic and mutation