Open Access

Gunter von Minckwitz, Hans-Peter Sinn, Günter Raab, Sibylle Loibl, Jens-Uwe Blohmer, Holger Eidtmann, Jörn Hilfrich, Elisabeth Merkle, Christian Jackisch, Serban D. Costa, Angelika Caputo, Manfred Kaufmann

• Introduction To investigate the predictive value of clinical and biological markers for a pathological complete remission after a preoperative dose-dense regimen of doxorubicin and docetaxel, with or without tamoxifen, in primary operable breast cancer. Methods Patients with a histologically confirmed diagnosis of previously untreated, operable, and measurable primary breast cancer (tumour (T), nodes (N) and metastases (M) score: T2-3(>= 3 cm) N0-2 M0) were treated in a prospectively randomised trial with four cycles of dose-dense (bi-weekly) doxorubicin and docetaxel (ddAT) chemotherapy, with or without tamoxifen, prior to surgery. Clinical and pathological parameters (menopausal status, clinical tumour size and nodal status, grade, and clinical response after two cycles) and a panel of biomarkers (oestrogen and progesterone receptors, Ki-67, human epidermal growth factor receptor 2 (HER2), p53, bcl-2, all detected by immunohistochemistry) were correlated with the detection of a pathological complete response (pCR). Results A pCR was observed in 9.7% in 248 patients randomised in the study and in 8.6% in the subset of 196 patients with available tumour tissue. Clinically negative axillary lymph nodes, poor tumour differentiation, negative oestrogen receptor status, negative progesterone receptor status, and loss of bcl-2 were significantly predictive for a pCR in a univariate logistic regression model, whereas in a multivariate analysis only the clinical nodal status and hormonal receptor status provided significantly independent information. Backward stepwise logistic regression revealed a response after two cycles, with hormone receptor status and lymph-node status as significant predictors. Patients with a low percentage of cells stained positive for Ki-67 showed a better response when treated with tamoxifen, whereas patients with a high percentage of Ki-67 positive cells did not have an additional benefit when treated with tamoxifen. Tumours overexpressing HER2 showed a similar response to that in HER2-negative patients when treated without tamoxifen, but when HER2-positive tumours were treated with tamoxifen, no pCR was observed. Conclusion Reliable prediction of a pathological complete response after preoperative chemotherapy is not possible with clinical and biological factors routinely determined before start of treatment. The response after two cycles of chemotherapy is a strong but dependent predictor. The only independent factor in this subset of patients was bcl-2. Trial registration number NCT00543829