In Silico Analysis of Breast Cancer Transcriptome Libraries Distinguish Tumor Subclasses.

Abstract

Abstract BackgroundBreast carcinoma is one of the most common causes of cancer related death worldwide. However marked differences in outcomes may reflect variation in diagnostic, staging and treatment. Serial Analysis of Gene Expression (SAGE) is a comprehensive profiling method that allows for global, unbiased and quantitative characterization of transcriptomes. A major advantage of SAGE is that once normalized it is possible to directly compare the levels of tags (short nucleotide sequences) generated by a single experiment with any other compatible available.MethodsTo gain an insight on the relationship between breast cancer transcriptomes and the disparate outcomes observed, we retrieve 18 SAGE libraries (4 Normal breast tissues, 11 primary breast tumors and 3 breast cancer metastatic tissues) from Cancer Genome Anatomy Project (CGAP). Data were analyzed by Correspondence Analysis (COA), Hierarchical clustering, Support Tree (ST) and Significance Analysis for Microarray (SAM) using TMEV software (tm4.org). Tags assignment to genes was performed by CGAP, and pathways analysis by FatiGO (babelomics.bioinfo.cipf.es).ResultsThe selection process to find SAGE tags that were consistently expressed in all normal, tumor and metastatic libraries resulted in 2,437 tags. COA shows clear separation between normal and metastatic libraries. These results were confirmed by Support Tree using the Pearson Correlation and Average Linkage (Fig 1). Interestingly, the 11 tumor SAGE libraries consisted of two heterogeneous groups, one similar to metastatic libraries (cases 1, 2, 8, 9, 10, 11) and the other to normal libraries (cases 3, 4, 5, 6, 12). Next, to identify SAGE tags differentially expressed between these two different groups of tumors, we performed SAM (delta value = 1.38, fold discovery rate = 0, 1001 unique permutations and fold change = 10). This approach revealed 59 (6%) tags differentially expressed across both tumor libraries. Among these 59 tags, 3 were up-regulated in tumors associated to normal libraries (normal-like tumors) and 56 were up-regulated in tumors associated to metastatic libraries (metastatic-like tumors). Next, tag to gene association identified 48 genes. Pathway analysis of these 48 genes revealed that tumors similar to normal libraries were characterized by activation of apoptosis and protein kinase pathways (JUND, GADD45B) and tumors similar to metastatic libraries were associated with response to extracellular stimulus, cell surface receptor linked signal transduction, Wnt receptor signaling pathway, cell cycle, phosphorylation, mRNA metabolic process, cell proliferation, regulation of cell proliferation, DNA replication, peptide metabolic process, signal peptide processing (NDUFB2, RPS27A, COX6B1, RBBP4, NDUFB9). These pathways may be implicated in outcomes to drug therapies.ConclusionsComprehensive analysis 2,437 tags from whole transcriptomes of primary breast cancer tissue compared with normal breast tissue and breast cancer metastatic tissues revealed that clinically disparate outcomes could be linked to a relatively small number of transcripts, since 6% (59 tags) were responsible of the differences across the normal-like and the tumor-like breast tumors.Supported FONDECYT 1080563 – Government of Chile Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 1165.