Abstract 3943: Affymetrix whole genome microarray analysis of 51 human tumor cell lines representative of 16 different tissues of origin: Unsupervised hierarchical clustering

The total mRNA for each sample of 51 human tumor cell lines representative of 16 different tissues of origin was split into 3 replicates, and analyzed against the entire human genome using standard Affx WT procedures. Approximately 50% of the samples (25 out of the 51 cell lines tested) exhibited low-level clustering related to their tissue of origin. The remaining 50% (26 out of the 51 cell lines tested) did not cluster with other samples of the same tissue of origin. These data reveal the importance of testing potential anticancer agents in multiple models representative of several different tumors of origin, as there is a 50% chance that the model chosen is not actually representative of the intended tissue of origin. This analysis also showed that the pancreatic cancer cell line CFPAC-1 did not cluster with any other cell line tested, revealing the unique genetic profile of this cell line. Interestingly, the reported lung cancer cell lines NCI-H69 and NCI-H82 clustered more closely with leukemic lines than with lung or any other solid tumor. This is particularly interesting as these lines are known to grow/behave more like a suspension culture than a monolayer. The NCI recently published its genetic analysis of their 60-panel, and they revealed that the MDA-MB-435 cell line, traditionally thought to be a breast cancer cell line, more closely resembled a melanoma line; hence, it was re-classified as a melanoma (likely a metastasized melanoma that was taken from the breast site). Our analysis reveals that another traditional breast cancer cell line, UISO-BCA-1, also clusters more closely with the melanomas (including the MDA-MB-435 cell line), suggesting that this cell line also may have been misclassified. Based on these data, we suggest that any potential anticancer agent showing activity in a particular cell line should be tested in other cell lines that cluster with the active line, and not merely in other lines supposedly representative of the same tissue of origin. Furthermore, in early stage testing, it would be more prudent to test several cell lines from different clusters, rather than several cell lines from different tissues of origin. It follows that by testing orphan drugs against several cell lines from each cluster, it would be possible to significantly narrow (and possibly identify), the likely drug target. 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 3943. doi:10.1158/1538-7445.AM2011-3943