PD03-04: SuperPathway Analyses of Luminal and Basaloid Breast Cancers from the Cancer Genome Atlas (TCGA) Program.

Abstract

Abstract The biological and clinical heterogeneity of breast cancer is clearly evident by its different intrinsic transcriptome subtypes. With exception of the HER2 subtype, pathways and signaling networks driving and distinguishing the other major breast cancer subtypes (basaloid, luminal-A, luminal-B) remain largely undefined. As of April 2011, >300 TCGA breast cancer samples have been characterized by both genomic copy number analysis and mRNA expression profiling (44% luminal-A, 26% luminal-B, 16% basaloid). As well, overall survival (OS) data are presently available on 104 luminal-A and 52 luminal-B cases, showing significantly poorer outcome for the latter of these hormonally driven subtypes (log rank p<0.05). Since the poorer outcome of luminal-B breast cancers may be due to their harboring more TP53 gene mutations (∼34%), we used a gene expression signature reported for TP53 mutated estrogen receptor-positive (ER+) breast cancers (Coutant et al., 2011), and confirmed that the TCGA luminal-B cases significantly overexpress this signature relative to luminal-A breast cases in which TP53 mutations are uncommon (6%). Curiously, within each luminal subtype, while this signature correlated with TP53 mutations it did not correlate with TP53 loss of heterozygosity (LOH), which was also higher in luminal-B (∼30%) relative to luminal-A (∼10%) cases. We used the network analysis tool PARADIGM (Vaske et al., Bioinformatics 26, i247-245) to integrate both DNA copy number and transcriptome data and infer pathway activity and interaction differences between luminal-A and luminal-B cases, and between the luminal and basaloid subtypes. This tool merges features derived from curated signal transduction, transcriptional and metabolic pathways into a Superimposed Pathway (SuperPathway) containing ∼3.1K unique activities, 1820 of which revealed significant differences among the breast cancer subtypes (Kruskal-Wallis test, Benjamin Hochberg FDR-corrected p <0.05). From the significant pathway activities differentiating basaloid from luminal-A (1399) or luminal-B (1122) subtypes, assessed using two independent approaches (functional enrichment/EASE scores, subnetwork analysis and hub interconnectivity >10 edges), higher FOXA1 (ER signaling) and lower HIF1A/ARNT transcription factor hub activities emerged as shared luminal differences relative to basaloid breast cancers. Of the 433 significant activities differentiating luminal-A from luminal-B subtypes, MYC/MAX, FOXM1, and MYB emerged as more active hubs in luminal-B, with normal TP53 function as a more active hub in luminal-A breast cancers. In sum, these TCGA analyses offer further insights into the signaling pathways differentiating basaloid from luminal breast cancers, and reveal major transcription factor hubs distinguishing the two common hormone dependent subtypes, luminal-A and luminal-B breast cancers. These findings suggest that SuperPathway analyses may inform therapeutic target opportunities and promote clinical development of breast cancer subtype specific treatment regimens. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr PD03-04.