Abstract 4035: Concurrent genes signatures for breast cancers with Han Chinese origin.

Abstract

Abstract We performed genome-wide characterization of Han Chinese breast cancer at molecular level by integrating two microarray technologies: comparative genomic hybridization (CGH) for DNA copy number variations (CNV) and expression arrays. Concurrent gains and losses from the same subject across genomic and transcriptional levels may be a better approach to explore potential biomarkers for breast cancer and to identify possible candidates for target therapy. Oncogenesis of breast cancer could originate from chromosome level manifesting as CNV and persist through transcription in gene expression profiles. Genes with coherent patterns at chromosomal and transcriptional level are more likely to serve as potential biomarkers for Han Chinese breast cancer and deserve meticulous evaluation. A total of 23 array CGH and 81 gene expression microarrays (21 samples with data from both platforms) were performed. Potential targets of cancer therapy were revealed by Genomic Identification of Significant Targets in Cancer (GISTIC) from 23 array CGH. Genes with coherent patterns between CNV and gene expression profiles were identified from 21 samples with both platforms assayed. We used these concurrent genes as well as genes with significant GISTIC scores to derive signatures associated with ER, HER2 and event-free survival. Distributions of signature genes were strongly associated with chromosomal locations: chromosome 16 for ER and 17 for HER2. 37 and 13 genes were differentially expressed between distinct ER and HER2 status and were used for classifier development in our 81 samples. Predictive accuracy as high as 90% for ER and 80% for HER2 was reported during leave-one-out cross-validation while much compromised performance was observed in an independent test dataset (GSE5460). Consensus genes between our dataset and 125 Chinese breast cancers (GSE5460) were identified, and a 28 genes and a 9 genes signature for ER and HER2 was developed in combined dataset of 206 microarrays with the best predictive accuracy of 92% and 94% for ER and HER2 respectively. Breast cancer risk predictive model was built based on supervised principle components from two genes (RWDD3 and ZBTB44) and distinct survival patterns were observed between high- and low-risk groups from a combined dataset of 408 microarrays (combination of our 81 samples with GSE20685). The risk score was significantly higher in breast cancers with recurrence, metastasis or mortality than those remained event-free (0.438 versus 0.17, p<0.001). In summary, we built a breast cancer risk predictive model and distinct survival patterns were observed in Han Chinese breast cancers. Citation Format: Chi-Cheng Huang, Shih-Hsin Tu, Eric Y. Chuang. Concurrent genes signatures for breast cancers with Han Chinese origin. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 4035. doi:10.1158/1538-7445.AM2013-4035