Survival of premenopausal women with metastatic breast cancer. Long-term follow-up of Eastern Cooperative Group and Cancer and Leukemia Group B studies.

In premenopausal women with metastatic breast cancer, differences in survival curves early during follow-up can be misleading. The authors therefore analyzed long-term survival in 378 patients, entered in three randomized trials, started between 1973 and 1978. Combined data from the three trials were used to increase the power for identifying prognostic variables. Cancer and Leukemia Group B (CALGB) trial 7382 randomized patients to oophorectomy plus either cyclophosphamide or combination chemotherapy or observation. Eastern Cooperative Oncology Group (ECOG) 2174 randomized patients who had not progressed 3 months after oophorectomy to combination chemotherapy or combination chemotherapy or observation. Trial ECOG 2177 randomized estrogen receptor (ER) positive or ER-unknown patients to oophorectomy plus combination chemotherapy or immediate combination chemotherapy, and ER-negative patients were directly assigned to combination chemotherapy. Hence ER-negative patients need not have been healthy enough to be randomized to oophorectomy. With only 14% of the patients still alive, median survival on the three studies was 30, 24, and 28 months. The median survival of individual treatments changed noticeably in ECOG 2174 and ECOG 2177 with long-term follow-up. At this time there are no differences in survival between randomized regimens in any of the three trials. In a multivariate model, factors associated with significantly poorer survival were visceral-dominant disease, nodal metastases, breast metastases, age younger than 45 years, ER negativity, and not receiving chemotherapy immediately after oophorectomy. This treatment difference was thus not due to imbalances in the prognostic variables used in the model, but it may be due to imbalances of unknown prognostic factors or differences in patient selection.