Our aim was to determine whether biological molecular markers can predict response to neoadjuvant chemoendocrine therapy in patients with early breast cancer. Ninety patients (median age 56 years; range, 28-69 years) with primary operable breast carcinoma were studied. They were treated with four 3-weekly cycles of chemotherapy with mitozantrone, methotrexate (+/- mitomycin C), and tamoxifen prior to surgery. Fine-needle aspiration was used to obtain samples from patients prior to therapy, and the following parameters were assessed: estrogen receptor (ER), progesterone receptor (PgR), p53, Ki67, Bcl-2, and c-erbB-2 measured by immunocytochemistry, and ploidy and S-phase fraction (SPF) by flow cytometry. The tumors of 78% of the subjects responded (complete response, 9%; partial response, 69%) and 22% did not (no change, 20%; progressive disease, 2%). Response rates according to disease stage and patient age were as follows: T1, 74%; T2, 79%; T3/T4, 78%; age </=50 years, 76%; >50, 79% (P = not significant). Response rates for other parameters were as follows: ER-positive, 82%, and -negative, 70%; PgR-positive, 86%, and -negative, 71%; p53-positive, 74%, and -negative, 81%; Bcl-2-positive, 85%, and -negative 61%; c-erbB-2-positive, 57%, and -negative, 93%; Ki67 high, 77%, and low, 81%; SPF high, 77%, and low, 77%; aneuploid, 71%; and diploid, 85%. Only the difference for c-erbB-2 was statistically significant (P = 0.007). A trend for higher response rates to neoadjuvant chemoendocrine therapy for tumors that were positive for ER, PgR, and Bcl-2 was observed but did not reach statistical significance. Tumors negative for c-erbB-2 had a higher response rate, which was statistically significant. In contrast, Ki67, ploidy, SPF, and p53 failed to predict for response.