Abstract 4418: Affymetrix whole genome microarray analysis of 24 human tumor xenograft models and the cell lines from which they were developed: Clustering of targets vs tumor tissue of origin

The NCI recently performed microarray expression analysis on its 60-panel of human tumor cell lines. This revealed important information, with some cell lines shown to be from a different tissue of origin than originally believed. Traditionally, potential anti-cancer agents have been evaluated in these in vitro models, and then moved into the corresponding in vivo xenograft model(s) based on the in vitro results. It has been shown that drugs which are effective in vitro are not necessarily effective in vivo and vice versa. Systematic microarray analysis of traditional xenograft models in conjunction with their in vitro counterparts has not been performed. The development of a human tumor xenograft in a mouse might be expected to lead to changes in gene expression, and this could account, in some instances, for the disconnect in results observed between in vitro and in vivo models. Our aim was to perform a genetic analysis against the entire human genome using 24 cell lines from 11 differing tissues of origin that were implanted into immune-deficient mice to establish a xenograft model for each. Once the tumors reached approximately 1 cm3 in size, the tumors were removed, cut into approx. 2-3 mm3 fragments, and an in vivo tumor passage was established. Microarray expression in fragments of those xenografted tumors was compared to microarray expression in the cell line from which they were developed. The total mRNA for each sample was split into 3 replicates, and analyzed against the entire human genome using standard Affx WT procedures. The results showed that over 60% (15 of 24) of the xenograft samples clustered with the cell line from which it was developed, whereas approximately 40% (9 of 24) did not, revealing that major changes in gene expression had occurred in 40% of these xenograft samples. Furthermore, when analyzed alone, these particular 24 cell line samples clustered according to their tissue of origin, whereas the tumor fragment samples did not appear to cluster. On the basis of these data we are currently performing the same analysis on an additional 25 tumor fragments and their corresponding matched cell lines to allow for a more accurate, in-depth cluster analysis. These data strongly suggest that although precedent exists to select in vitro models on the basis of their tissue of origin, no such precedent exists for in vivo models. In vivo models should be more carefully selected to ensure that the model chosen is still representative of the tissue to be tested. It follows that a drug candidate effective in a particular in vitro model might be expected to show activity in other in vitro lines from the same tissue of origin. However, a drug candidate effective in a particular in vivo model representative of a tissue of origin should not be expected to show efficacy in other in vivo models representing the same tissue type. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 4418. doi:10.1158/1538-7445.AM2011-4418