A microRNA Expression Profile Consisting of 15 microRNAs, Including miR205, Is Associated with Poor Prognosis in Breast Cancer.

Abstract

Abstract Introduction: MicroRNAs (miRNA) are a class of non-coding RNAs able to regulate gene expression at the post-transcriptional level. In breast cancer, levels of specific miRNAs differ between malignant and normal breast tissue and are able to classify tumors according to clinicopathologic variables. This highlights the potential of miRNAs as novel prognostic and/or predictive indicators. In this study we sought for miRNAs denoting poor prognosis in breast cancer.Materials and methods: 377 miRNAs were profiled in 70 breast tumor samples using the human MicroRNA A Array Set version 2.0 (Applied Biosystems). All miRNAs with Ct-values less than 35 in 25% of the samples were included for analysis. Data normalization was performed relative to the median miRNA expression level per sample and expression values were log2-transformed. Principal component analysis (PCA) was performed to identify metagenes of miRNAs associated with clinicopathological variables (TNM-status, tumor stage, histological grade, ER-, PR-, ErbB2- and P53-status) and prognostic/predictive gene signature-classifications (wound healing response, invasiveness gene signature, 70-gene prognostic signature, genomic grade index, recurrence score, HOXB13/IL17RB-expression ratio and the molecular breast cancer subtypes). Results were validated by investigating relationships between the expression of a relevant miRNA target gene signature and clinicopathological variables in 5 publicly available breast cancer gene expression data sets.Results: Using PCA-analysis we identified a metagene significantly associated with the Luminal B phenotype, an elevated genomic grade index, an elevated recurrence score, an activated wound healing response, the invasiveness gene signature and poor prognosis according to the 70-gene prognostic profile (range Rs: 0.325-0.372; P<0.05). Using multivariate regression analysis we demonstrated that this metagene was also associated with loss of PR expression (Hazard Ratio (HR): 9.00; P=0.0001) and the presence of metastases at time of diagnosis (HR: 18.9; P=0.0002). MiRNAs significantly (FDR<0.1) associated with this metagene were: miR205, miR548C-5p, miR548D-5p, miR187, miR671-3p, miR492, miR365, miR499-5p, miR548D-3p, miR615-3p, miR193A, miR589, miR660, miR219-1-3p, miR888 (range Rs: -0.415 – 0.456). At the gene expression level, a miR205 target gene signature, composed of 173 unique genes, was associated with relapse-free survival (range HR: 1.225 – 1.387; P<0.01) and distant metastasis-free survival (HR: 3.950; P=0.006) in respectively 3 and 1 independent breast cancer data set(s).Discussion: Through principal component analysis we identified a miRNA signature associated with poor prognosis in breast cancer. This signature was composed of 15 individual miRNAs, including miR205 which is known for its role in epithelial-to-mesenchymal transition and the induction of stem cell characteristics. The association of a miR205 target gene signature with relapse-free and distant metastases-free survival confirms our results on the gene expression level. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 4062.