Do we need biomarkers to predict the benefit of adding adjuvant taxanes for treatment of breast cancer?

The current understanding of the molecular biology of breast cancer presents an extremely complex portrait of the disease. Based on this knowledge, considerable efforts are being made to identify biomarkers that will predict the response to a specific treatment while minimizing the risk of unnecessary side effects. In breast cancer, the Ki67 index has been associated with poor prognosis and might play a relevant role in predicting benefit from adjuvant docetaxel, as observed in the article accompanying this editorial. Taxanes are one of the most active cytotoxic agents for breast cancer. However, the role of taxane-based chemotherapy as adjuvant treatment of early breast cancer remains controversial in some subsets of patients. For this reason, the Ki67 index might help to better define the group of patients who could have the optimal benefit.

higher chemosensitivity in breast cancer. However, assessment of the Ki67 index is still a matter of debate because of the lack of consensus in defi ning an optimal cutoff level.
In the previous issue of Breast Cancer Research, Jacquemier and colleagues [1] retrospectively assessed the prognostic and predictive value of subtyping breast cancer by immunohistochemistry and other multiple biomarkers in the PACS 01 trial. In this study, 1,999 patients with node-positive breast cancer were randomly assigned to receive six cycles of fl uorouracil, epirubicin, and cyclophosphamide (FEC) or a sequential regimen of three cycles of FEC followed by three cycles of docetaxel [2]. Th e addition of docetaxel signifi cantly improved the 5-year rates of disease-free survival (DFS) (73.2% versus 78.4%; P = 0.012) and overall survival (86.7% versus 90.7%; P = 0.017).
In the article, the expression of 34 selected proteins, which included immunohistochemical determination of the Ki67 index and status of ER, progesterone receptor (PR), and HER2 as well as the evaluation of 30 additional proteins by tissue microarrays, was centrally analyzed in 1,099 tumor samples retrospectively. Th e authors concluded that high Ki67 index levels, defi ned as at least 20% of tumor cells displaying positive nuclear staining, were associated with shorter DFS but greater docetaxel benefi t in patients with node-positive breast cancer treated with adjuvant anthracycline-based therapy. In addition, it was found that the luminal A subtype did not benefi t from docetaxel treatment.
Th is study provides important insights into the biology and response to therapy of breast tumors. However, at least three considerations should be raised. First, although several studies have reported that higher Ki67 index levels predict better clinical and pathological response to neoadjuvant chemotherapy, the correlation between Ki67 index and the specifi c treatment benefi t has not, to the best of our knowledge, been rigorously evaluated. Viale and colleagues [3] failed to demonstrate

Abstract
The current understanding of the molecular biology of breast cancer presents an extremely complex portrait of the disease. Based on this knowledge, considerable eff orts are being made to identify biomarkers that will predict the response to a specifi c treatment while minimizing the risk of unnecessary side eff ects. In breast cancer, the Ki67 index has been associated with poor prognosis and might play a relevant role in predicting benefi t from adjuvant docetaxel, as observed in the article accompanying this editorial. Taxanes are one of the most active cytotoxic agents for breast cancer. However, the role of taxane-based chemotherapy as adjuvant treatment of early breast cancer remains controversial in some subsets of patients. For this reason, the Ki67 index might help to better defi ne the group of patients who could have the optimal benefi t.

© 2010 BioMed Central Ltd
Do we need biomarkers to predict the benefi t of adding adjuvant taxanes for treatment of breast cancer?
that Ki67 index levels predict which patients may benefi t from adding a non-anthracycline adjuvant chemotherapy regimen to endocrine therapy. More recently, adjuvant assessment tools for the prediction of treatment benefi t, including Oncotype Dx (Genomic Health, Redwood City, CA, USA) and MammaPrint (Agendia, Irvine, CA, USA), have been used to aid clinical decision making [4,5]. Oncotype Dx measures the expression of 21 genes, including proliferation-related genes such as MKi67, as major determinants in calculating an Oncotype Dx score [6]. With this molecular platform, both node-negative and node-positive patients with a high recurrence score showed a clear benefi t from combined cyclophosphamide, methotrexate, and 5-fl uorouracil (CMF) and anthracycline-based chemotherapy. Despite previously reported data to the contrary, the Ki67 index could play a relevant role in predicting benefi t from adjuvant chemotherapy. Th erefore, the Ki67 index seems to be a predictor of higher chemosensitivity in patients with breast cancer but is unable to predict the benefi t of a specifi c treatment.
Second, although state-of-the-art intrinsic subtyping of breast tumors is based on molecular profi ling of gene expression data from microarrays, these techniques are not ready for use in daily clinical practice. As a result, a combination of immunohistochemical surrogate markers, taking into account ER and PR status, HER2 status, histological grade, and Ki67 index, is being used for 'molecular subtyping' . However, there is currently no consensus on how to establish a single immuno histochemical defi ni tion of each molecular subtype. 'Molecular sub typing' has been used to predict the benefi t of adding adjuvant taxanes in patients with ER-positive breast cancer. A meta-analysis of 13 randomized trials evaluated the effi cacy of including taxanes to anthracycline-based regi mens. Th e addition of taxanes resulted in absolute 5-year risk reductions of 5% for DFS and 3% for overall survival in patients with high-risk early breast cancer. Th is benefi t was independent of ER expression and type of taxane [7]. However, unplanned retrospective analyses performed in each study have reported contradictory results. In relation to paclitaxel, the CALGB 9344 trial showed that, in comparison with patients with ERnegative or HER2-positive breast cancer, those with HER2-negative, ER-positive, and node-positive breast cancer obtained little benefi t from the administration of paclitaxel after adjuvant doxorubicin-based therapy [8]. However, other studies have not confi rmed these results [9,10]. On the other hand, the effi cacy of docetaxelcontaining adjuvant regi mens was independent of ER expression in two random ized trials (PACS 01 and BCIRG 001) [2,11]. Th erefore, the benefi t of taxanes appears to be independent of ER expression.
Currently, it is well established that ER-positive breast cancers are highly heterogeneous. Th e luminal B subtype has a cell proliferation signature that includes expression of the MKi67 gene. Consequently, this subtype expresses high Ki67 index levels and has a signifi cantly worse prognosis than luminal A tumors. Th erefore, luminal B tumors may benefi t from additional systemic therapy [12][13][14]. However, because luminal A tumors were observed to benefi t from paclitaxel in the GEICAM 9906 study, it is unclear whether the benefi t of taxanes is exclusively confi ned to patients with luminal B tumors. We cannot rule out the possibility that luminal A tumors benefi t from adjuvant chemotherapy (a lack of benefi t from taxanes is diff erent than that from chemotherapy).
Th ird, the limitations of retrospective studies are well known. However, we must also be cautious when interpreting the results of subgroup analyses. Wang and colleagues [15] defi ned a subgroup analysis as 'any evaluation of treatment eff ects for a specifi c endpoint in subgroups of patients defi ned by baseline characteristics' . Th is analysis is usually performed to evaluate the consistency of a trial's conclusions among patient subgroups included in the study [15]. A subgroup analysis can potentially support or alter our clinical decisions or do both. For this reason, we must take into account that inappropriate subgroup analysis can lead to invalid results. Consequently, an unplanned subgroup analysis should be considered exploratory and hypothesisgenerat ing only.
In conclusion, it is clear that the Ki67 index is a prognostic factor and a powerful predictor of higher chemosensitivity in patients with breast cancer. However, there are not enough data to establish a relationship between the Ki67 index and a therapy-specifi c benefi t. As for the clinical implications of this study, 'molecular subtyping' and the Ki67 index will help to recommend the use, but not the type, of adjuvant chemotherapy.