Individualized Therapy of Breast Cancer: Are we there Yet?

The history of medicine is intricately intertwined with technologic developments in diagnostic methods to define diseases ever more narrowly and to predict clinical outcome with or without particular therapies with increasing precision. Every new analytical method that can be applied to human tissues offers a new opportunity to advance this gradual progress towards individualized treatment recommendations. High-throughput gene-expression analysis that allows the semi-quantitative assessment of thousands of mRNA species in a small tumor biopsy is no exception. In recent years, this technology has made important contributions to our view of breast cancer as a disease and also led to the development of several novel diagnostic assays, at least one of which is increasingly used in routine practice. Invasive ductal carcinoma, the most common type of breast cancer, used to be considered a single disease entity with heterogeneous expression of various molecular markers including estrogen receptors (ERs). Gene-expression profiling revealed large-scale molecular differences between ER-positive and ER-negative cancers; ER itself represents only the tip of a vast iceberg of molecular abnormalities that distinguish these two types of breast cancers [1]. As a consequence of these discoveries, it is more appropriate to consider breast cancer as a collection of several distinct diseases, not unlike the term pneumonia, which includes a number of different inflammatory conditions of the lung caused by different insults that require different treatments. This conceptual shift has important clinical ramifications. It implies that different clinical trials and treatment strategies may be appropriate for the distinct molecular types of breast cancer because these have different clinical courses and different sensitivity to existing treatment modalities [2,3]. It will be important to start to develop new drugs and treatment strategies separately for different subsets of breast cancer. Similarly, different molecular predictors of outcome (i.e., prognostic or predictive biomarkers) will be needed for the different types of breast cancers. One size fits all, poorly. Returning to the analogy of pneumonia, no one today would consider developing a drug to treat all pneumonias in general, because very different drugs are needed to combat bacterial pneumonia compared with viral or fungal pneumonias or noninfectious inflammatory diseases of the lung. Indeed, funding agencies began to recognize this fundamental shift in our understanding of breast cancer and started to solicit breast cancer molecular subtype-specific grant submissions [101].

Another, more subtle but important impact of high-dimensional genomic data on breast cancer research was that it brought into the forefront the need to use a combination of variables, genes or clinical features, to predict disease outcome. It is a mathematical fact that combinations of independent variables, that are each associated to various degrees with a particular outcome, will yield a more accurate multivariate-prediction model than any single variable (i.e., gene) alone. The clinical value of considering several factors together in order to make the best estimate of prognosis has been recognized empirically for several decades. An example is the tumor–node–metastasis (TNM) prognostic staging of breast cancer that creates clinical stages I, II and III based on the size of the primary tumor (T), the number of lymph nodes involved (N) and presence or absence of distant metastasis (M) [4]. However, this historical staging system does not combine these important prognostic variables into a mathematically correct multivariate prediction model, but relies on empirical groupings of these variables. The TNM system does capture prognostic information but may not use these variables optimally to create the most accurate prognostic estimate for an individual. Physicians increasingly use a mathematically more correct, multivariate, freely available, web-based prognostic model, Adjuvant! Online [102], that relies on many of the same clinical variables as TNM but provides an individualized risk prediction [102]. Similarly, research emphasis is shifting from finding individual genes as best predictors of prognosis (or response to therapy) to examining combinations of genes as markers [5].

The first novel multigene diagnostic assay that has entered widespread clinical use illustrates several of the aforementioned points. Oncotype Dx^® (Genomic Health, CA, USA) measures the expression of 21 genes including 16 cancer-related genes and five controls. The cancer-related genes fall into categories that collectively assess the proliferative activity, the ER and human epidermal growth factor receptor-2 (HER-2) status of the cancer as well as some other genes. This multivariate gene expression-based prognostic predictor was developed for a particular subset of breast cancers, ER-positive cancers [6]. The Oncotype Dx recurrence score, which is a weighted sum of the expression of the 16 cancer-related genes, does not predict recurrence accurately in ER-negative cancers, but it fills an important diagnostic niche in ER-positive cancers. Many patients with early-stage, ER-positive breast cancer have prolonged cancer-free survival with appropriate surgery and adjuvant (i.e., postoperative) therapy with anti-estrogens or drugs that inhibit estrogen production. However, a subset of ER-positive cancers have poor outcome despite these treatments and could benefit from inclusion of chemotherapy in their treatment. Until Oncotype Dx, there was no standardized and widely accepted test to define this high-risk population requiring adjuvant chemotherapy among the ER-positive cancers. Oncotype Dx is now used to predict the risk of recurrence of ER-positive, early-stage breast cancers treated with endocrine therapy, low-risk individuals may safely avoid adjuvant chemotherapy, whereas high-risk patients need to consider further treatment with chemotherapy [7]. This assay will also allow the conduct of more efficient clinical trials for ER-positive breast cancers in the future [8]. Numerous other multigene predictors of prognosis, sensitivity to endocrine therapy or chemotherapy have also been developed [5]. Each of these represents important avenues of research, but not all are equally ready for clinical use owing to the variable strength and extent of supporting data. Investigators also started to combine these various predictors into a single multipurpose assay that uses information from different sets of genes to simultaneously but separately estimate prognosis and response to various treatment modalities using gene-expression information from a needle biopsy [9].

The ultimate dream of personalized treatment, building a treatment regimen that is tailored to the molecular abnormalities of a particular cancer, is yet to be realized. However, small but clinically important advances have been made and are exemplified by the various novel, first-generation, multigene diagnostic assays that can contribute to better medical decision making for at least some patients. Equally importantly, the technological and conceptual framework is in place to make further progress. Comprehensive molecular analysis of small-needle biopsy specimens of cancer is already feasible today using gene-expression profiling, high resolution comparative genomic hybridization and gene sequencing. The major molecular types of breast cancer have been defined and the need to develop drugs and conduct clinical trials separately for these different cancers is increasingly recognized. The already existing high-dimensional molecular databases of human breast cancer represent a rich treasure trove to discover the next generation of drug targets. A particularly appealing aspect of this approach is that the therapeutic target identification process also defines methods to select patients for future therapy. Last but not least, the biomarker research community and diagnostic companies have started to adopt many of the methodologies that have been used successfully by the pharmaceutical industry, including clinical trial-based approaches to demonstrate clinical value for markers and cost–effectiveness analyses to show monetary value and education programs for physicians to teach them how to use novel assays. These important developments in recent years make the prospect of further advances in personalized medicine for breast cancer bright.

Financial & competing interest disclosure

LP holds shares in Nuvera Biosciences Inc. and received honoraria from Bristol-Myers-Squibb, Roche Pharmaceuticals and Dako Inc. for ad hoc consulting services. Research in the author‘s laboratory is supported by the Breast Cancer Research Foundation, The Susan Komen Foundation and an award from the American Society of Clinical Oncology. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.

Additional information

Funding

LP holds shares in Nuvera Biosciences Inc. and received honoraria from Bristol-Myers-Squibb, Roche Pharmaceuticals and Dako Inc. for ad hoc consulting services. Research in the author‘s laboratory is supported by the Breast Cancer Research Foundation, The Susan Komen Foundation and an award from the American Society of Clinical Oncology. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed. No writing assistance was utilized in the production of this manuscript.