In the metastatic context, selection of specific breast cancer treatment currently depends on hormone-receptor and HER2 status. For those patients who are non-responsive to endocrine therapy or have tumors negative for hormone receptors, standard chemotherapy remains the most common treatment option with single agent regimens usually consisting of either anthracyclines, taxanes, cyclophosphamide, fluorouracile, capecitabine, vinorelbine, or gemcitabine [1, 2]. Unfortunately, there is no specific recommendation at present for second-line treatment or further chemotherapy as no particular regimen has been shown to offer greater efficacy . In fact, from the 60% of patients with early-stage breast cancer that will receive adjuvant chemotherapy, only 2–15% will ultimately derive benefit from treatment, while all treated patients will be exposed to toxic side-effects .
The primary objective of pharmacogenomics is to develop markers able to address specific aspects of response and/or toxicity and help in the individualization of breast cancer therapy. Biomarkers can be broadly categorized either as prognostic when solely associated with clinical outcome, and predictive when associated with the effectiveness of a specific drug. A prognostic marker is a unique molecular feature or set of features assembled as a signature, which can separate populations of patients based on disease outcome in the absence of treatment or despite a non-specific treatment. A predictive marker is, on the other hand, a unique molecular feature or signature of features that can separate patient populations based on clinical outcome derived from a specific targeted therapy. When a predictive marker has been properly validated, it can help to identify patients most likely to expect to benefit, or be less susceptible to suffer side effects, from a particular therapy.
The quest for reliable predictive biomarkers for cytotoxic agents has been, and will certainly remain, a long and challenging enterprise. As we gain a better understanding of the weaknesses of the available methodologies, it is becoming increasingly evident that each step in the analysis process is critical with regard to accuracy, reproducibility, and predictive value of new markers or signatures . In order to minimize inaccuracies, complementary techniques have been selected in parallel to assess the usefulness of these biomarkers in predicting the response of an individual patient to a specific therapy.
Three different strategies were used to identify markers or signatures for each cytotoxic agent used in the Center of Excellence (COE) breast cancer retrospective cohorts.
Based on the concept that potential biomarkers have a better chance of being linked to a clinical response, a set of biomarkers were identified as either targets of a particular cytotoxic agent or determinants of its metabolism (Table 5.1). Gene copy number, or protein expression, of some of these selected markers, was evaluated using fluorescent in situ hybridization (FISH) and immunohistochemistry (IHC), respectively.
Using an mRNA expression profiling microarray-based assay (WG-DASL; Whole Genome cDNA-mediated Annealing, Selection extension and Ligation), genes ...