Although cytotoxic chemotherapy is widely used in advanced breast cancer, there are no powerful predictors for the therapy response. Because topoisomerase IIalpha (Topo IIalpha) is the molecular target for the anthracycline class of anti-cancer drugs, we compared the immunocytochemical assay of Topo IIalpha with other biomarkers in the prediction of clinical response to Topo II inhibitor chemotherapy. Fifty-five patients with advanced breast cancer were treated with a single cytotoxic drug, Topo II-inhibitor, epirubicin (30 mg m(-2) weekly up to 1000 mg m(-2)), as first line cytotoxic chemotherapy. Objective response to treatment was analysed according to UICC criteria. The predictive value of Topo IIalpha expression, c-erbB2 oncoprotein, p53 tumour-suppressor protein, oestrogen (ER) and progesterone receptor (PR), S-phase fraction and DNA ploidy were analysed from representative formalin-fixed paraffin-embedded primary tumour samples. The proportion of Topo IIalpha-positive cells (Topo IIalpha index) failed to predict response to epirubicin therapy. Mean Topo IIalpha scores in 29 responding patients were similar when compared with those with no change in disease progression (n = 13) and those with progressive disease (n = 13) (14.9% +/- 11.4% vs 15.5% +/- 7.6% vs 17.3% +/- 13.2%, not significant). Among the other biomarkers tested, overexpression of c-erbB2 oncoprotein and hormone receptor negativity were significantly associated with poor response. Response rate in patients with c-erbB2-overexpressing tumours was 32% compared with 65% in patients with no c-erbB2 overexpression (P = 0.0058). Accordingly, the response rate for ER-positive patients was 67% compared with 26% in ER-negative patients (P = 0.0021). Although both negative ER status and c-erbB2 overexpression are associated with high Topo IIalpha expression in breast cancer, step-wise logistic regression analysis showed that ER and c-erbB2 were associated with therapy response independent of Topo IIalpha expression. Histological grade, p53, DNA-ploidy, tumour proliferation rate (S-phase fraction), stage of the disease at diagnosis, age of the patient, previous anti-oestrogen therapy or site of metastasis did not predict the response to epirubicin therapy. In conclusion, despite extensive in vitro evidence, expression of Topo IIalpha is unlikely to predict the response to Topo II inhibitor chemotherapy in advanced breast cancer. Among the prognostic biomarkers, overexpression of c-erbB2 oncogene and negative ER may have predictive value in epirubicin therapy in patients with advanced breast cancer.