Abstract 1473: Are the genomic gene expression profiles maintained between the original donor and patient-derived xenograft tumors

Abstract

Abstract INTRODUCTION: Patient-derived xenograft model (PDX) that transplant patient tumors into nude mice are now an accepted model to be used for drug development, and even for guiding therapy. It has been reported that PDX tumors retain the histologic characteristics, tumor heterogeneity, and maintain the metastatic potential of the original donor tumor. Therefore, PDX has been expected to be a potential tool for predicting responses to specific therapies, and large amounts of funding and labor are invested in generating PDX from tumors obtained from patient samples. Recently, however, there have been disagreements in the literature whether PDX tumors maintain the transcriptomics gene expression profile of the original donor tumor. METHODS: 2 datasets (Uronis (PLoS One. 2012;7(6):e38422), and Julien (Clin Cancer Res. 2012 Oct 1;18(19):5314-28)) were identified from the GEO database that contained gene expression data from colorectal cancer tumor samples and from xenografts of the same tumors transplanted into nude mice or rats. A series of quality control tests were performed to ensure only high quality arrays were included in downstream analyses. Differentially expressed genes (FDR < 0.01) in each dataset were identified comparing patient donor tumor and PDX tumor. RESULTS: There were 19 original tumors and 27 PDX in the Uronis dataset and 40 original tumors and 75 PDX in the Julien dataset. In both datasets, there was a general under-expression of genes in PDX tumors compared with original donor tumors. Hierarchical clustering of the Uronis dataset showed two main clusters - the donor tumors clustered together and the PDX tumors clustered separately, except for one original tumor sample that clustered together with the PDX. There were a total of 5234 genes differentially expressed between the donor and PDX tumors. In the Julien dataset, the PDX tumors also clustered separately from the original tumors. There were also 2 main clusters seen in hierarchical cluster analysis, with most of the original tumors clustering together except for eight, and all the PDX clustering together including the former eight original tumors. There were a total of 1661 differentially expressed genes, with 913 genes differentially expressed in both datasets (p = 7.4e-151). CONCLUSIONS: Original donor tumors and matched PDX tumors cluster separately in hierarchical cluster analysis. These results suggest that PDX transcriptomic gene expression profiles are more similar among PDX tumors that originated from different tumors than from the original tumor used to create the corresponding PDX. Thus we recommend using the PDX model with caution due to the significant dissimilarity of gene expression profiles between PDX and matched donor tumors. Citation Format: Leopoldo J. Fernandez, Aaron R. Wolen, Amy L. Olex, Mikhail Dozmorov, David A. Fenstermacher, Kazuaki Takabe. Are the genomic gene expression profiles maintained between the original donor and patient-derived xenograft tumors. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 1473. doi:10.1158/1538-7445.AM2015-1473