|Home | About | Journals | Submit | Contact Us | Français|
Breast cancer clinically represents a heterogeneous disease. Over the last decades, the integration of prognostic and predictive markers in treatment decisions has led to a more individualized and optimized therapy. While prognosis describes the risk of disease recurrence and disease-related death after diagnosis without the influence of therapy, prediction illustrates the probability of efficacy or response of a specific therapeutic measure. The substantial decline in breast cancer mortality seen over the last 20 years is primarily due to the delivery of adjuvant systemic therapy. It is important that clinical decisions are made to minimize overtreatment, under-treatment, and incorrect treatment. Improved understanding of breast cancer biology together with the utilization of classical biomarkers and the identification of new markers or profiles is increasingly defining who should receive cancer therapy and what therapy offers the best efficacy. The molecular targets as the prerequisite for successful concepts of specific therapies like anti-estrogens, antibodies, or small molecules, have therefore high clinical value in regards to prognosis as well as prediction.
The development of specific systemic treatment options in early breast cancer have led to a substantial decline in breast cancer mortality over the last 20 years . The observed advances are based on the identification of patient subgroups who are in need of treatment and the definition of markers which allow the prediction of the efficacy of certain treatment measures. The best established prognostic markers for breast cancer comprise tumor size, nodal status, metastases, histological tumor type, grading, and age, as well as peritumoral lymphovascular invasion (LVI) . Table Table11 summarizes the established markers and lists the grade of recommendation according to the AGO (Arbeitsgemeinschaft Gynákologische Onkologie) Breast Committee guidelines for the diagnosis and treatment of patients with primary and metastatic breast cancer [2, 3].
Tumor size is a strong and independent prognosticator for breast cancer, even after 20 years of follow-up [4, 5] and exhibits a positive correlation to the axillary lymph node status . However, efficacy of chemotherapy is independent from tumor size.
For early breast cancer, the axillary lymph node status still represents the most important prognostic factor. Node-positive patients exhibit a 4–8 times higher mortality than node-negative ones , and the number of metastatic lymph nodes correlates directly to the risk of recurrence and death . However, the efficacy of adjuvant therapy is not influenced by this parameter. According to international and national guidelines, patients with lymph node-positive breast cancer will be normally treated with adjuvant chemotherapy, independent of receptor status, since they represent the group of patients with the highest benefit from this treatment strategy . For lymph node-negative patients, additional prognostic and predictive markers have to be considered for adequate adjuvant treatment decisions and are still a pertinent issue of ongoing translational research and clinical trials (table (table2).2). Approximately 30% of node-negative patients will need chemotherapy, but the identification of this subgroup and a clear discrimination against the 70% of patients who are sufficiently treated by surgery, radiation, and endocrine treatment, is one of the most difficult questions.
The prognostic impact of the histological subtype is limited. Most breast cancers belong to the ductal-invasive and lobular-invasive type which show no significant difference in the clinical course of disease and are therefore no discriminator for treatment decisions. However, some rare breast cancers like the tubular, mucinous, and invasive cribriform type do not metastasize frequently and have therefore an excellent prognosis .
Histological grade is another prognostic factor for early breast cancer with a strong independent value [4, 5]. The validity of grading has been compromised by the inter-observer reproducibility. The switch from the traditional Scarff, Bloom, Richardson nuclear system to the modern Elston and Ellis grading system has increased reproducibility significantly . Especially in node-negative patients, higher grade (especially G3) is correlated with an adverse course of disease and endocrine insensitivity, and represents the strongest prognostic factor for this subgroup of patients. These patients should be treated with adjuvant chemotherapy.
Very young age at the time of diagnosis is associated with poor survival. Women younger than 35–40 years exhibit a larger likelihood to develop local recurrences as well as distant métastases in comparison to older patients [4, 5, 9]. This might be due to the fact that they present more frequently with larger tumor size, affected lymph nodes, estrogen receptor (ER) negativity and Her2/neu overexpression [9, 10]. In comparison to patients 40–49 years of age, women aged < 35 and 35–39 years have a 2.2 and 1.4 higher risk of death, respectively . Young age still qualifies patients as being at high risk with the consequence of an adequate systemic treatment with chemotherapy. In this case, the risk of recurrence and death is comparable to the risk of patients older than 40 years . Even if actual studies including gene signatures cannot support the independence of age as a prognostic factor in multivariate analyses , the predictive value of the meno-pausal status with regard to endocrine therapy options is without doubt important. Surgical or pharmacological ovarian suppression as well as the contraindication for the use of aromatase inhibitors in premenopausal patients underline the importance of this parameter in the daily use for treatment decisions.
LVI is an additional prognostic factor [3, 12]. On the one hand, LVI is regarded as an additional risk factor for axillary lymph node involvement; on the other hand, it is supposed to have a special impact for node-negative patients even if the evidence is quite inhomogeneous. Some studies found an up to 60% higher mortality rate for node-negative patients with a LV1 status in comparison to LV0 tumors .
The determination of hormone receptor expression is a widely accepted standard procedure for breast pathology. Interestingly, ER and progesterone receptor (PR) have both prognostic and predictive value, even if the predictive power is much stronger and consecutively more commonly used. Expression of ER reflects a favorable tumor biology, even if its prognostic value is inadequate. The absence of ER predicts earlier recurrence, while ER-positive patients have a constant recurrence rate after 5 years .
Of highest clinical importance is the predictive value of ER, since its presence correlates with the benefit of adjuvant endocrine therapy . In this regard, it is of interest that the level of ER expression directly correlates with the probability of treatment response and efficacy, especially for the treatment with tamoxifen . Furthermore, there is growing evidence for a significantly better response to chemotherapy for hormone receptor-negative patients .
The independent value of PR expression is an ongoing discussion. While the prognostic power is inconclusive and not confirmed, even the predictive significance is being debated. The EBCTCG metaanalysis could not confirm the earlier postulated independent predictive power of PR for treatment response, even if clinical observations support an increased benefit for ER- and PR-positive tumors in comparison to ER-positive/PR-negative tumors .
The methodology for ER and PR measurement has changed over the last decades and is currently based on im-munohistochemistry. The percentage of stained tumor cells or, in addition, a score as a product of percentage of stained cells and staining intensity (Remmele-Stegner score, 0–12) is widely used . While the international St. Gallen Consensus Conference defined any staining as the threshold for predicted endocrine responsiveness of the tumor , the AGO guidelines specify responsiveness with a staining of ≥ 1% .
The impact of Her2/neu on breast cancer biology, the clear definition of a molecular target, and the development of a highly effective therapeutic option in the application of the monoclonal antibody trastuzumab form one of the most dramatic changes in breast cancer therapy over the last decades and have opened up the field of personalized, targeted treatment besides endocrine therapy. Her2/neu as a member of the transmembraneous Her family is overexpressed in 15–20% of tumors, mainly due to amplification of the Her2/neu gene. Overexpression is strongly correlated with aggressive tumor type, downregulation of hormone receptors, and induced proliferation, with consecutive decreased overall survival, although these observations are still being debated for node-negative patients without adjuvant therapy. While some trials were not able to detect a biological impact of Her2/neu , there is growing evidence for an independent prognostic power of Her2/neu in this specific subgroup of patients [3, 20]. However, Her2/neu is an important and routinely used predictive marker. Besides retrospective analyses which showed a reduced efficacy with regard to certain endocrine (tamoxifen) and cytotoxic (cyclophosphamide, methotrexate, 5-fluorouracil) strategies , the predictive power of Her2/neu is based on the application of the monoclonal antibody trastuzumab. Only patients with Her2/neu overexpression, determined by immunohistochemistry (> 30% intense and complete staining) or in situ hybridization (≥ 6.0 signals per nucleus by single color or signal ratio ≥ 2.2 for Her2/neu versus centromere 17 for dual color), will have a clinical benefit from a trastzumab-based therapy with a consistently confirmed 50% reduced risk of recurrence [3, 21, 22].
A broad variety of new biomarkers with potential prognostic or predictive value are discussed in the literature, but only a few have proven to represent an evidence-based gain for daily use.
The tumor-associated fibrinolytic factors urokinase-type plasminogen activator (uPA) and its inhibitor type 1 (PAI-1) are important promoters of tumor invasion and metastasis. The uPA/PAI-1 assay is based on a standardized enzyme-linked immunosorbent assay (ELISA) to measure the protein levels in fresh tumor samples. Single clinical trials as well a recently published metaanalysis with > 8,000 patients confirmed the prognostic value of uPA/PAI-1 . Node-negative patients with low uPA/PAI-1 levels have a very low risk of recurrence, and in the final analysis of the multicentric prospective Chemo N0 trail a 10-year survival rate of nearly 90% without any adjuvant treatment was observed . Besides retrospective analyses showing a clear benefit for adjuvant chemotherapy in patients with high versus low uPA/PAI levels, the prospectively designed international multicentric NNBC-3 study will address the question of treatment optimization for node-negative patients with a taxane/anthracycline-based chemotherapy regimen with regard to uPA/PAI expression. Based on the well established standardization and the solid scientific evidence, uPA/PAI-1 detection is recommended by the AGO Breast Committee (AGO LOE la, GR A, +) as well as by the American Society of Clinical Oncology (ASCO) guidelines, especially for the group of G2 node-negative patient, as a tool for treatment decisions .
Proliferation has been recognized for a long time as an important prognosticator in breast cancer. Ki-67 represents a strong proliferation marker and has been gaining more interest over the last years. Patients with highly proliferating tumors exhibit an increased benefit from adjuvant therapy in comparison to patients with low or intermediate proliferation activity . In retrospective analyses, higher values of Ki-67 were associated with adverse prognostic factors such as Her2/neu expression, higher grading, or LVI, and with worse disease-free survival with a hazard ratio (HR) of 1.8 in comparison to tumors with low Ki-67 expression . Since prospective validation is still missing, a general recommendation for Ki-67 determination cannot be given. However, this factor represents an additional potential prognostic marker in selected cases and is gaining more importance due to its role in defining the different intrinsic subtypes of breast cancer .
Over the last years, molecular characterization of breast cancer has gained more and more importance [26, 27]. The profound analysis of gene expression profiles has led to the definition of 5 different molecular intrinsic subtypes of breast cancer: ER-positive/luminal A and B, basal-like, ErbB2-positive, and normal breast, which are associated with differences in clinical outcome. The intrinsic subtypes as distinct entities were found to have a significant impact on recurrence-free survival in untreated patients and remained significant in multivariate analysis incorporating standard prognostic factors such as ER status, histological grade, tumor size, and lymph node status. The basal-like subtype – which is mostly G3-differentiated, expressing C5/6 cytokeratins, and negative for ER, PR, and Her2/neu – is characterized by an unfavorable prognosis and represents a potential predictor for specific adjuvant systemic treatment, which is currently under investigation . According to the St. Gallen Consensus 2011, the intrinsic subtypes can be defined by immunohisto-chemistry, in analogy to the classification recommended by Cheang et al.  using ER, PR, Her2/neu, and Ki-67, respectively: i) Luminal A = ER- and/or PR-positive, Her2/neu-neg-ative, Ki-67 low (< 14%); ii) Luminal B (Her2/neu-negative) = ER- and/or PR-positive, Her2/neu-negative, Ki-67 high (< 14%); iii) Luminal B (Her2/neu-positive) = ER- and/or PR-positive, Her2/neu overexpression/amplification, Ki-67 low or high; iv) Her2/neu-positive (not luminal) = ER- and PR-negative, Her2/neu overexpression/amplification; v) Basal-like, triple-negative (ductal) = ER- and PR-negative, Her2/ neu-negative.
The significant discrepancies between the clinical and molecular classification become apparent for the so-called triple-negative breast cancers (ER-, PR-, Her2/neu-negative) which are defined immunohistochemically. This subgroup is not identical to the molecularly defined basal-like subtype and underlines the fact that ER and Her2/neu status are not accurate surrogates for the true intrinsic subtype status. As a consequence, the optimal classification system for breast cancer subtypes to guide therapeutic decision-making has to yet be defined. Nevertheless, the triple-negative as well as the basal-like subtypes are characterized by an adverse course of disease and the need for adjuvant chemotherapy, even though the optimal regimens (platinum-based chemotherapy, PARP inhibitors) are the current focus of ongoing clinical trials .
The detection of disseminating tumor cells (DTCs) is an additional field of growing interest in breast cancer. The risk of distant recurrence is biologically based on minimal residual disease (MRD) which can be detected by the identification of DTCs in the bone marrow or circulating tumor cells (CTCs) in the blood. The relevance of DTCs as an independent prognostic parameter was demonstrated in several studies and in 1 metaanalysis . However, so far, no routine detection method has been established since the therapeutic consequences are unclear. The detection of CTCs in peripheral blood is technically difficult, but a standardized FDA-approved assay is currently available. CTCs represent a validated prognostic marker in the metastasized situation and correlate with significantly shorter survival . In the adjuvant situation, several studies were able to identify a prognostic value for CTCs . However, since there is no proof of a clinical benefit resulting from therapeutic measures undertaken in response to the existence of CTCs, the detection of MRD cannot currently be recommended for routine use .
Multigene analyses offer the possibility of simultaneous investigation of multiple tumor-relevant pathways with the goal of identifying prognostic and predictive gene expression signatures . Different test platforms are used to detect gene profiles. Quantitative reverse transcription polymerase chain reaction (qRT-PCR) and cDNA microarray are complex methods which also require complex mathematical algorithms to avoid false associations. The feasibility of these approaches was demonstrated in several retrospective studies, but the required prospective data are missing so far.
The Oncotype DX™ (Genomic Health Inc., Redwood City, CA, USA) assay is based on qRT-PCR to avoid the problem of fresh frozen tumor tissue, and uses readily available paraffin-embedded tumor blocks. A set of 250 genes was primarily analyzed in different study populations including the NSABP B-20 trial. The combination of gene expression could be linked to clinical outcome, and by applying a continuous variable algorithm, 3 risk categories could be defined – low, intermediate, and high-risk – for developing distant métastases at 10 years. Finally, a 21-gene set (16 cancer-related, 5 reference genes) could be identified for the optimal discrimination of the 3 risk categories, and this set was validated in the NSABP B-14 trial in a population of ER-positive, node-negative patients who underwent endocrine therapy with tamoxifen . The most dominant genes in this set can be grouped to proliferation (e.g. Ki-67), Her2/neu, and ER-related genes. A recurrence score is calculated with low score (< 18), intermediate score (> 18 and < 30), and high score (> 30). In the validation population of the NSABP B-14 trial, these score translated into a risk for distant recurrence of 7% (low), 14% (intermediate), and 31% (high), respectively [35, 36]. The molecular classification seems to better predict clinical outcome in comparison to conventional prognostic markers. A recently published retrospective study supported these findings and was able to show that up to 50% of patients who were classified with classical parameters to be at high risk could be downgraded to low risk by the utilization of Oncotype DX . Besides the validation data for tamoxifen treatment, retrospective analyses (including NSABP B-20) also revealed that node-negative patients had no benefit from an additional chemotherapy in the case of a low recurrence score . However, until now, the AGO guidelines have not recommended the usage of this assay since prospective data are missing . Several large prospectively designed multi-center trials (TAILORx, Plan B) are ongoing to evaluate Oncotype DX for treatment decision, especially with regard to chemotherapy.
A different approach for multigene analyses is the utilization of fresh frozen tissue by applying cDNA arrays. The most advanced and analyzed assay is the MammaPrint™ (Tromms-dorff GmbH & Co. KG, Alsdorf, Germany) using a 70-gene signature. Starting from 25,000 genes, the Amsterdam group analyzed the clinical outcome (endpoint: distant-free survival) of 78 patients < 55 years of age with node-negative tumors less than 5 cm which were ER-positive or -negative. The retrospectively chosen validation cohort proofed the 70-gene signature to be a potent discriminator for a good versus bad prognosis with a 10-year survival of 85 vs. 51%, respectively . A further multicentre validation study of 300 untreated primary breast cancer patients exposed the potential prognostic significance of the assay by showing a hazard ratio of 4.6 vs. 2.1 for patients with a bad prognosis in comparison to a good prognosis signature . Even if these data for multigene signatures are promising, the clinical value is so far not proven since no prospective data exist. The MammaPrint assay is currently the subject of a large prospective international trial (MINDACT of the BIG study group) to validate its prognostic power. Therefore, the current AGO guidelines do not recommend the clinical utilization of this test outside of clinical trials .
In the daily clinical setting, the established classical prognostic and predictive biomarkers still represent the backbone for treatment decisions with the highest clinical value. Some innovative markers, like uPA/PA-1, also represent solid and highly validated tools to guide treatment decisions, especially in certain subgroups of patients with early breast cancer, even if long-term data are mostly unavailable. It is of big importance to note that the innovative multigene approaches potentially will open a new chapter in breast cancer therapy but currently fail to reach the required validation level for routine clinical use. Besides all the discussed factors, additional risk modifiers such as lifestyle (obesity, alcohol consumption) have proven influence on disease outcome in breast cancer patients and should as a consequence be considered and verbalized in the communication with the patient.