Stromal changes have been the focus of numerous research publications and have led to insights in both tumor development and promising new avenues for treatment. In order to study molecular changes in stroma from tissue samples it is recommended to separate tumor tissue from stromal tissue. One such context is mouse tumor xenograft models where tumors, particularly metastatic tumors, can be small and difficult to separate from the host tissue. In our research we compared qualitatively the ability of RNA-seq and microarray data to detect tumor (human) and stromal (mouse) expression from mixed samples in terms of cross-alignment and cross-hybridization. Samples were analyzed using HumanWG-6_V3_0_R1 and MouseWG-6_V2_0_R0 Expression BeadChips and the GenomeAnalyzer IIe (Illumina, Inc.). Samples consisted of total RNA from normal mouse lung from NOD/SCID gamma mice and total RNA from MDA-MB-231 breast cancer cells combined in fixed proportions in triplicate. We define a gene which cross-hybridizes from mouse to human as one which exists in the set defined by (B U C U D) - A, where A, B, C, and D are defined as follows. A is the set of all genes detected when pure mouse RNA was hybridized onto mouse chips (or aligned to the mouse genome); B is the set of all genes detected when 25% human/75% mouse RNA was hybridized onto mouse chips (or aligned to the mouse genome); C is the set of all genes detected when 50% human/50% mouse RNA was hybridized onto mouse chips (or aligned to the mouse genome); D is the set of all genes detected when 75% human/25% mouse RNA was hybridized onto mouse chips (or aligned to the mouse genome). A gene that cross-hybridizes from human to mouse is defined analogously. Our results show that observed levels of cross-hybridization are quite low (5.32% of human probes detected in mouse, 3.48% of mouse probes detected in human). The observed levels of cross-alignment are practically comparable to the levels of cross-hybridization (6.50% of human genes detected in mouse, 2.27% of mouse genes detected in human). However, there are genes and pathways of considerable interest to oncology researchers which show significant cross-hybridization/cross-alignment and, as such, their presence/absence or level of expression in tumor tissue versus stromal tissue cannot be determined using the platforms from this study. Cross-hybridizing/cross-aligning genes in our studies include PDGF, b-Raf, Beta-catenin, erbB2, NF-kB, MDM2, Claudin, VEGF-R, Notch2, Cyclin B, HSP90 and Ubiquitin. Biological pathways significantly enriched for genes which show cross-hybridization/cross-alignment include TGF-mediated regulation of cell proliferation, TGF/WNT and cytoskeletal remodeling, regulation of EMT, hedgehod signaling, and the FGFR signaling pathway. Which platform to use with mixed tissues from xenografts - microarrays or NGS - appears to be primarily a question of cost and specific genes of interest.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 2994. doi:1538-7445.AM2012-2994