REVIEW ARTICLE Int J Clin Oncol (2002) 7:245–253 © The Japan Society of Clinical Oncology 2002 Angelo Di Leo · Fatima Cardoso · Virginie Durbecq Rosa Giuliani · Max Mano · Gul Atalay · Denis Larsimont Christos Sotiriou · Laura Biganzoli · Martine J. Piccart Predictive molecular markers in the adjuvant therapy of breast cancer: state of the art in the year 2002 Received: February 28, 2002 A. Di Leo (*) · F. Cardoso · V. Durbecq · R. Giuliani · M. Mano · G. Atalay · C. Sotiriou · L. Biganzoli · M. J. Piccart Department of Chemotherapy, Jules Bordet Institute, Brussels, rue Heger-Bordet 1, 1000 Brussels, Belgium Tel. +32-2-541-31-80; Fax +32-2-541-30-90 e-mail: Angelo.Dileo@bordet.be D. Larsimont Department of Pathology, Jules Bordet Institute, Brussels, Belgium Abstract In the present article, an extensive review of the current knowledge regarding predictive molecular markers in breast cancer is presented. The main emphasis has been given to the adjuvant therapy setting, because of the lack of well-designed studies evaluating predictive markers in the metastatic setting. In the first part of the article some gen- eral concepts have been summarized, mainly to emphasize the difference between prognostic and predictive markers. In the second part of the article, studies evaluating mole- cular markers with predictive value for the most commonly used drugs or regimens (i.e., cyclophosphamide, metho- trexate, 5-fluorouracil [CMF], anthracyclines, taxanes, and herceptin) have been reviewed and discussed. An attempt has been made to provide for each single predictive marker the currently available level of evidence, to allow the reader to have immediate information regarding the impact of the evaluated marker in daily clinical practice. In the last part of the article, the most important findings have been summa- rized and future avenues of research have been outlined. Key words Breast cancer · Adjuvant therapy · Molecular markers · Predictive markers Introduction The results of important and large randomized clinical trials and the conclusions of the Oxford meta-analysis have had a major impact on treatment recommendations for patients with early breast cancer 1–7 (R. Peto 2000, Fifth main meet- ing of the Early Breast Cancer Trialist’s Collaborative Group, unpublished work). Today, adjuvant medical therapy is offered to the vast majority of patients with node- negative breast cancer. 1,4–6 Indeed, the international panel of experts on adjuvant therapy met again in St. Gallen in February 2001, and concluded that only node-negative pa- tients aged 35 years or more, with tumors of small size (2 cm), well-differentiated (histological grade 1), and posi- tive for both the estrogen (ER) and the progesterone (PgR) receptors, might be spared adjuvant chemotherapy because of their minimal risk of relapse. Adjuvant chemotherapy should at least be offered to all other patients, particularly those with endocrine non-responsive disease (i.e., ER- and PgR-negative). 8 The Oxford 2000 overview, based on data from 56 trials and 28 000 women, concluded that, after stan- dardization for age and time since randomization, with the use of polychemotherapy, proportional reductions in risk of recurrence and death are similar for women with node-negative and node-positive disease (R. Peto 2000, unpublished). According to these guidelines, most patients with early breast cancer receive adjuvant treatment, and the main challenge has become the identification of predictive factors that may help in selecting the optimal therapeutic strategy for individual patients. The accomplishment of this ambi- tious aim could translate into a substantial increase in the absolute benefit associated with adjuvant therapy. Predictive factors: general concepts The main differences between predictive and prognostic factors are illustrated in Fig. 1. While a prognostic factor influences disease outcome whichever adjuvant therapy is used (Fig. 1A), a predictive factor will interfere with disease outcome only when a specific treatment is given (Fig. 1B,C). Figure 1B shows how the outcome of treatment A is positively influenced by the predictive factor under study, which, in contrast, has no impact on the efficacy of treat- ment B. A predictive factor might also differentially influ- ence the outcome of two different adjuvant therapies, as described in Fig. 1C. This latter scenario would be the ideal