Background: Liver is the most common metastatic site in advanced colorectal cancer. Most patients with colorectal cancer liver metastasis (CRLM) could not benefit from the current treatment. Patient-derived xenograft (PDX) models with definite molecular signature are attractive preclinical models and essential for development of novel drugs. Aim: This study was designed to focus on the establishment, characterization of pathologic and molecular features of PDX models. Furthermore, we also validate potential therapeutic targets and explore novel drug therapies guided by genotyping or expression profiling, leading to potential implications for precision medicine. Methods: CRLM PDX models were established and elucidated their possible implications for preclinical research and personalized treatment from their fidelity of clinicopathologic characteristics, genomic landscape, and antitumor activities of novel targeted drugs. Response biomarkers were also explored. Results: A total of 56 PDX models from CRLM were successfully established (transplantation success 76.7%, 56/73). The transplantation rate was higher than that of primary specimens (61.5%, 16/26). No differences were observed between latency period and characteristics except the level of CEA. Along with the passaging, latency period became shorter and shorter. PDXs from CRLM recapitulated the pathologic, genetic, and protein properties of corresponding parental tumors. Frequent altered genes showed high consistency compared with patients' genomic alterations, and were enriched in MAPK, ErbB, cell cycle, focal adhesion, adherence junction pathways. Several potential drug targets, such as KRAS, HER2 and FGFR2, were selected and validated by corresponding inhibitors. In addition, PDX models could also used for patients with no druggable alterations identified to screen the efficient regimen. Conclusion: In this study, we have successfully established and validated a large panel of molecularly annotated CRLM platforms for preclinical evaluation of novel therapeutics and biomarker discovery.
