Patients and Methods: Mitoxantrone, methotrexate (with or without mitomycin), and tamoxifen chemoendocrine therapy was administered to 158 patients before surgery. Clinical response was assessed after four cycles of treatment. Fine-needle aspiration cytology for estrogen receptor (ER), progesterone receptor (PSR), c-erbB-2, p53, bcl-2, Ki67, S-phase fraction (SPF), and ploidy were performed pretreatment and repeated on day 10 or day 21 after the first cycle of treatment.

Results: Good clinical response (GCR, defined as complete response or minimal residual disease) wets achieved in 31% of patients (49 of 158). Tumor size, nodal disease, response, ER, pgR, c-erbB-2, p53, bcl-2, Ki67, SPF, and ploidy were analyzed as predictors of survival. By univariate analysis, node-positive disease (P = .05), lack of ER (P < .05) and PgR (P < .05), and failure to attain GCR (P = .008) were associated with a significantly increased risk of relapse. A significantly increased risk of death was associated with node-positive disease (P = .02), lack of ER expression (P = .04), and failure to attain GCR, By multivariate analysis, GCR was an independent predictor for survival (P = .05). ER expression (P = .03), absence of c-erbB-2 (P = .03), and a decrease in Ki67 on day 10 or day 21 of the first cycle (P < .05) significantly predicted for subsequent GCR.

Conclusion: Molecular markers may be used to predict the likelihood of achieving GCR, which seems to be a valid surrogate marker for survival. (C) 1999 by American Society of Clinical Oncology.