Gene expression signatures in breast cancer

Gene expression analysis has depicted the striking molecular heterogeneity in human breast cancer. DNA microarray analyses of fresh-frozen tissue identified several so-called intrinsic subtypes, such as luminal A, luminal B, ERBB2-like (HER2) and basal-like. Alternatively, gene expression signatures were used to distinguish patients with low risk or high risk for distant metastasis. To improve feasibility it is important that these gene expression analyses can be performed utilizing formalin-fixed paraffin-embedded tissue. The aim of this evidence-based review was the prognostic significance of gene expression signatures in early breast cancer. A systematic search in PubMed and a manual search were carried out to review relevant articles published between 2000 and 2013. Key words used were "gene-expression", "intrinsic subtypes", "Endopredict", "Mammaprint", "Oncotype DX", "PAM50", "level of evidence" and "immune system" in combination with "prognosis" and "breast cancer". To avoid an inaccurate risk classification it is mandatory that meticulous clinical and analytical validation of these tests is performed and that a high level of evidence (LoE) is achieved. The commercially available gene expression signatures of the first generation, such as EndopredictA (R) (LoE I), MammaprintA (R) (LoE II), Oncotype DXA (R) (LoE I) and PAM50 (LoE II) are currently rated as +/- by the Working Group on Gynecological Oncology (AGO) suggesting that these tests may be performed only in individual cases and that a general recommendation cannot be given. These signatures can be used for risk classification especially in estrogen receptor (ER) positive patients. In contrast, signatures of the second generation use immune cell-related transcripts to assess prognosis better especially in ER negative or HER2 positive rapidly proliferating breast cancer. Gene expression signatures with a high LoE can result in an improved assessment of prognosis in early breast cancer.