Still Waiting After 110 Years: The Optimal Use of Ovarian Ablation As Adjuvant Therapy for Breast Cancer

Abstract

Although Schinzinger first suggested that oophorectomy might be used to treat breast cancer in 1889, it is Sir George Beatson who is associated with its establishment as an effective therapeutic modality based on his seminal 1896 Lancet publication describing successful treatment of young women with advanced breast cancer with bilateral oophorectomy. Studies of surgical oophorectomy were followed by those assessing ovarian irradiation, leading ultimately to the 1948 launch at the Christie Hospital (Manchester, United Kingdom) of the first randomized trial of ovarian ablation (OA) using irradiation versus no OA for premenopausal women with operable breast cancer. By the 1960s, many considered OA to be an effective adjuvant modality in breast cancer, but interest subsided with the demonstration of the efficacy of adjuvant chemotherapy and the belief that OA failed to alter survival in premenopausal women. This view was challenged by the serial reports of the Early Breast Cancer Trialists’ Collaborative Group showing that OA or ovarian suppression (OS) offered an unequivocal survival benefit compared with no therapy for women under 50 years of age. In addition, the ovarian suppressive or ablative effects of chemotherapy in young women were recognized, which raised the possibility that the benefits of chemotherapy might be mediated in part by indirect effects on the ovary leading to estrogen deprivation. These observations led to a generation of randomized trials assessing the efficacy of OA/OS by surgery, radiotherapy, or luteinizing hormone–releasing hormone agonists compared with chemotherapy or in addition to chemotherapy. In this issue, Ejlertsen et al report the long-term outcome of patients treated in such a trial, the Danish Breast Cancer Cooperative Group (DBCG) Trial 89B. This study randomly assigned 762 premenopausal women with steroid hormone receptor–positive early breast cancer deemed at high risk for recurrence (axillary nodal involvement or tumor size 5 cm) to receive pelvic irradiation as a form of OA/OS or nine courses of cyclophosphamide, methotrexate, and fluorouracil (CMF) intravenously every 3 weeks. Both therapies were well tolerated, and there was no apparent difference in overall survival after a median follow-up time of 10.5 years and 358 first events (hazard ratio 1.11; 95% CI, 0.88 to 1.42). The unique strengths of this study include its maturity, exclusion of women with known steroid receptor–negative breast cancer who are unlikely to benefit from endocrine therapy, focus on an essentially node-positive population at high risk for recurrence, and uniform approach to therapy. Potential limitations include the use of a chemotherapy regimen (CMF) that some might consider substandard in the era of taxanes and anthracyclines, failure to incorporate tamoxifen, and elements of the statistical design and power. The DBCG 89B investigators used a noninferiority statistical design to test the hypothesis that ovarian irradiation was not inferior to chemotherapy. Active-control noninferiority trials are often used when the choice of a placebo-controlled study is considered unethical. However, this type of trial design is challenging because of the need to define an acceptable margin of inferiority to reduce the risk of introducing a potentially suboptimal treatment into clinical practice while maintaining a reasonable sample size. The predefined sample size of 750 patients chosen for DBCG 89B permitted detection of a 25% relative decrease in disease-free survival with pelvic irradiation as clinically unacceptable using twosided testing with 5% and 20%. The observed unadjusted hazard ratio estimate for disease-free survival of 0.99 (95% CI, 0.81 to 1.22; P .95) rejects the null hypothesis of inequality and supports the alternative hypothesis of equality. However, the upper boundaries of the CIs do not rule out a possible 22% greater disease-free survival or a 42% greater overall survival benefit with CMF, which raises the question of whether the DBCG 89B trial was adequately powered to address a noninferiority question in a clinically meaningful way. Nonetheless, the conclusion of Ejlertsen et al that radiationinduced OA/OS and CMF have a similar effect on disease-free and overall survival for premenopausal women with receptor-positive breast cancer is likely to be correct. This study joins four other published studies comparing OA/OS by surgery, goserelin, or leuprolide with variations of CMF that have shown similar results for women with hormone-responsive disease. DBCG 89B is the only study that exclusively used irradiation to induce OA/OS and excluded women with known hormone receptor–negative disease (although 22% of the enrolled subjects had unknown receptor status), thereby limiting inclusion of patients expected to gain nothing from OA/OS and derive greater benefit from chemotherapy. A tragedy of all of these trials, as well as concurrent trials that tested chemotherapy followed by OA/OS, is that they failed to incorporate tamoxifen because of the prevailing belief at the time that it was not effective in premenopausal women with JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 24 NUMBER 31 NOVEMBER 1 2006