Abstract P4-12-01: Molecular features and survival outcomes of the intrinsic subtypes of breast cancer based on HER2 gene amplification

Abstract

Abstract Background: HER2-amplified (HER2+) breast cancer is biologically heterogeneous and all the intrinsic molecular subtypes (Luminal A, Luminal B, HER2-enriched and Basal-like) can be identified. However, the molecular and outcome differences of the intrinsic subtypes based on HER2 status have not been thoroughly studied. This is important since HER2 is considered an oncogene involved in the activation of various signal transduction pathways, as well as a biomarker of poor prognosis (in the absence of anti-HER2 targeting). Methods: To compare molecular features between clinically HER2+ and HER2-negative tumors, we interrogated The Cancer Genome Atlas (TCGA) publicly available dataset of 825 primary breast cancers with at least one of following data types: mRNA expression (17,784 genes), protein expression (171 proteins and phospho-proteins), DNA methylation status (574 probes), miRNA expression (306 transcripts) and whole exome somatic mutations. Two-class unpaired Significant Analyses of Microarrays were used to identify significant biomarker associations with a False Discovery Rate of 0%. For survival associations, we interrogated the METABRIC DNA copy number/gene expression-based dataset (Curtis et al. Nature) composed of 1,971 primary breast tumors with long-term clinical follow-up (and no adjuvant anti-HER2 therapy). Multivariable Cox models were used to test the prognostic significance of each variable. The research based 50-gene PAM50 model was used to classify tumors into the different intrinsic subtypes. Results: In both datasets combined (n = 2,225), HER2+ disease showed an enrichment for HER2-enriched tumors (47.0% vs. 7.1%) and a decrease in Luminal A tumors (7.3% vs. 39.0%) compared to HER2-negative disease (p<0.001). In the TCGA dataset, the percentage of HER2+ tumors within HER2-enriched (n = 55), Luminal B (n = 122), Luminal A (n = 223) and Basal-like (n = 95) subtypes were 70.9%, 16.4%, 6.3% and 2.7%, respectively. Within each intrinsic subtype, only between 13 to 44 genes (0.07% to 0.25% of all genes evaluated) were found more expressed in HER2+ tumors compared to HER2-negative tumors. The vast majority (77.5%) of these significant genes are located on the 17q12 chromosomal amplicon, such as HER2 and GRB7. Interestingly, luminal- or proliferation-related genes were not found differentially expressed when all HER2+ tumors were compared against all HER2-negative tumors within a given subtype. Similarly, only 6 to 7 proteins (ERBB2, pERBB2, EGFR, EGFR_pY1068, EGFR_pY992, RPS6KB1 and ACACA), mostly located in the 17q12 amplicon (except EGFR), were found differentially expressed between HER2+ and HER2-negative tumors within a given subtype. Minimal changes were also noted when DNA methylation patterns, miRNA gene expression and somatic mutations were evaluated. Finally, no additional prognostic value was observed with the addition of HER2 status to intrinsic subtype. Conclusions: When the intrinsic subtypes are taken into account, HER2 amplification does not translate into large changes in the activation of downstream signaling pathways or worse patient survival outcomes. These results also suggest that the potential responses to anti-HER2 therapy on HER2-amplified tumor cells depend in part upon their intrinsic tumor profile. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P4-12-01.