Meta-analysis of trials comparing toremifene with tamoxifen and factors predicting outcome of antiestrogen therapy in postmenopausal women with breast cancer

Meta-analysis of all clinical data was conducted to compare toremifene 40-6 0 mg/day (TOR) with tamoxifen 20-40 mg/day (TAM) in postmenopausal women wi th estrogen receptor (ER) positive or ER unknown advanced breast cancer and assess factors predicting treatment outcome. Data from five randomized par allel group studies (all studies) were combined. Efficacy variables were th e response rate in all studies and also the time to treatment failure and s urvival in the three major studies (pivotal studies). Of the 1421 patients, 725 received TOR and 696 TAM. Response rates were 24. 0 and 25.3, respectively (p = 0.675) with 95 confidence interval (95 CI) fo r the difference -5.3 to 3.4. Of the 1157 patients in the pivotal studies, 75 had progressed and 50 expired. Median treatment times were 4.9 months in TOR and 5.3 months in TAM groups (p = 0.762, hazard ratio 0.98 with 95 CI 0.87-1.11). Median survival times were 31.0 (TOR) and 33.1 (TAM) months (p = 0.758, hazard ratio 0.98 with 95 CI 0.83-1.15). All results are consisten t with the criteria of statistical equivalence between TOR and TAM. More pa tients in TAM (20) than in TOR (14, p = 0.007) discontinued the treatment p rematurely but overall the treatments were well tolerated. As the treatment s were equally effective all data were analyzed together for predictive fac tors. High tumor ER concentration, long disease free time, soft tissue meta stases, few metastatic sites, and good performance status all independently predicted longer survival (p < 0.001). Previous adjuvant tamoxifen predict ed shorter survival (p = 0.008). Objective response to treatment or disease stabilization for at least 12 months both predicted prolonged survival (p = 0.001). TOR 60 mg/day and TAM are equally effective and well tolerated in the treat ment of advanced breast cancer in postmenopausal women. Probability of surv ival may be predicted based on patient characteristics and on the initial r esponse to the treatment.