Résumé : PURPOSE: To evaluate the predictive value of HER-2 in a population of advanced breast cancer patients randomly treated either with single-agent doxorubicin (A) or with single-agent docetaxel (T). EXPERIMENTAL DESIGN: Patients from this study participated in a phase III clinical trial in which doxorubicin or docetaxel was administered for advanced disease. HER-2 was evaluated by IHC. In all positive cases, FISH was used to confirm the HER-2 positive status. The different cohorts of patients identified by HER-2 were examined to assess a possible relationship between HER-2 status and treatment effect. RESULTS: Tumor samples were available for 176 of the 326 patients entered in the clinical trial (54%). HER-2 positivity was observed in 20% of the study population. A statistically significant interaction was found between response rates to the study drugs and HER-2 status, with HER-2 positive patients deriving the highest benefit from the use of T (odds ratio for HER-2 positive patients treated with T = 3.12 (95% CI 1.11-8.80), p = 0.03). The interaction between HER-2 and response rates to A and T was also confirmed by a multivariate analysis. No statistically significant interaction was found between HER-2 and drugs efficacy evaluated in terms of time to progression and overall survival, although in the HER-2 negative cohort A was at least as effective as T in term of overall survival. CONCLUSIONS: These results suggest that in HER-2 positive breast cancer patients docetaxel might be more active than doxorubicin, while in HER-2 negative patients doxorubicin might be at least as effective as docetaxel. Although the present results cannot have an impact on current practice, they allow us to formulate the hypothesis that HER-2 positive breast cancer is a heterogeneous disease with regard to sensitivity to anthracyclines and taxanes, and that this might be dependent upon other molecular markers including the p-53 and topoisomerase II alpha genes. This hypothesis is currently being tested prospectively in two different 'bench to bed-side' clinical trials.