Abstract LB-042: Cancer cell intrinsic gene expression signatures minimize the confounding effects of intratumoral heterogeneity in colorectal cancer patient classification

Abstract

Abstract The application of transcriptional gene signatures to stratify tumors into prognostic and predictive subtypes has evolved rapidly since the original landmark studies demonstrated the clinical utility of this approach in breast cancer. While the number of published signatures continues to increase, the likelihood of individual signatures achieving clinical utility has remained very low, with some estimates putting this figure below 1%. Transcriptional intratumoral heterogeneity (ITH), resulting from variation in stromal components at different regions of a tumor, has been shown to undermine molecular stratification of colorectal cancer (CRC) patients into prognostic and predictive subgroups. The degree of gene expression changes associated with variation in tumor microenvironment (TME) content may even mask the relatively more subtle changes associated with genetic variability and heterogeneity within the tumor epithelium. This variation in TME content, and subsequent ITH, at different regions of the tumor has long been recognised; however, in the era of precision medicine, the implications of such microenvironmental alterations in confounding patient classification are of increased importance. To examine this issue, a panel of clinically relevant prognostic and predictive transcriptional gene signatures were selected in order to assess the potential for discordant classification of patient samples based on the tumoral region profiled. Using these clinically relevant CRC gene expression signatures, we assessed the susceptibility of each signature to the confounding effects of ITH using gene expression microarray data obtained from multiple tumor regions of a cohort of 24 patients, including central tumor, the tumor invasive front and lymph node metastasis. Consensus annotated gene lists for each of the gene signatures were used, followed by in silico analysis within our multiregional clinical dataset. Sample clustering alongside correlative assessment revealed variation in the ability of each signature to cluster samples according to patient-of-origin rather than region-of-origin within the multiregion dataset. Signatures focused on cancer-cell intrinsic gene expression were found to produce more clinically useful, patient-centred classifiers, as exemplified by the recently developed CRC intrinsic signature (CRIS), which robustly clustered samples by patient rather than sample region. These findings highlight the potential of cancer-cell intrinsic signatures to reliably stratify CRC patients by minimising the confounding effects of ITH. In conclusion, our data support the clinical evaluation of signatures based on intrinsic gene expression, such as CRIS, which are unaffected by the confounding variable of transcriptional ITH and therefore have the potential to be used clinically to inform precision medicine decisions, ultimately leading to improved patient outcomes. Citation Format: Philip D. Dunne, Paul O'Reilly, Aideen Roddy, Matthew Alderdice, Susan Richman, Tim Maughan, Simon McDade, Patrick Johnston, Daniel Longley, Elaine Kay, Darragh McArt, Mark Lawler. Cancer cell intrinsic gene expression signatures minimize the confounding effects of intratumoral heterogeneity in colorectal cancer patient classification [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr LB-042. doi:10.1158/1538-7445.AM2017-LB-042