Patient Eligibility and Entry Procedures
Women with primary operable HER2-negative breast cancer diagnosed by means of a core needle biopsy were eligible for participation in the study. Patients were required to have a palpable primary tumor at least 2.0 cm in diameter in the breast, as assessed by physical examination, and to be classified as having tumor stage T1c to T3, nodal stage N0 to N2a, and metastasis stage M0. Other key eligibility criteria were an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1 (with 0 indicating that the patient is fully active and able to carry on all predisease activities without restriction and 1 indicating that the patient is restricted in physically strenuous activity but is ambulatory and able to carry out work of a light or sedentary nature, such as light housework or office work) and normal left ventricular ejection fraction, determined by multiple-gated acquisition scanning or echocardiography within 3 months before study entry. Before random assignment, patients were required to have core-biopsy material harvested for correlative science studies.
The assignment to treatment groups was balanced according to age at entry (≤49 or ≥50 years), clinical tumor size (2.0 to 4 cm or ≥4.1 cm), hormone-receptor status (estrogen-receptor–positive, progesterone-receptor–positive, or both vs. estrogen-receptor–negative and progesterone-receptor–negative), and clinical nodal status (negative vs. positive). Randomization was performed within these strata, with the use of a biased-coin approach to ensure balanced treatment assignments within an institution.
The protocol was approved by the central institutional review board of the National Cancer Institute (NCI) and by the human investigations committee or institutional review board at each participating site, each of which has approval for human subjects research from the Department of Health and Human Services. Written informed consent was obtained from all participants. The first author wrote the first draft of the manuscript, and all the authors contributed to subsequent drafts and made the decision to submit the manuscript for publication. A group of NSABP authors ensured the fidelity of the study to the protocol. F. Hoffmann–La Roche, Genentech USA, and Eli Lilly provided partial support for this trial and were provided an opportunity to review the manuscript before submission for publication. The NSABP restricts the access of sponsors to outcomes data until the time of publication. The study had no additional commercial support, and no person other than the authors contributed to the content of the manuscript. The protocol, including the statistical analysis plan, is available with the full text of this article at NEJM.org.
Eligible patients were randomly assigned to one of three neoadjuvant chemotherapy regimens: four cycles of docetaxel (100 mg per square meter of body-surface area, administered intravenously on day 1 of the cycle) every 3 weeks, followed by four cycles of doxorubicin–cyclophosphamide (60 mg and 600 mg per square meter, respectively, administered intravenously every 3 weeks) (docetaxel group); capecitabine (825 mg per square meter, administered orally twice a day on days 1 through 14) added to docetaxel (75 mg per square meter, administered intravenously on day 1), followed by doxorubicin–cyclophosphamide (docetaxel–cape cit abine group); or gemcitabine (1000 mg per square meter, administered intravenously on days 1 and 8) added to docetaxel (75 mg per square meter, administered intravenously on day 1), followed by doxorubicin–cyclophosphamide (docetaxel–gemcitabine group) (see Fig. S1 in the Supplementary Appendix
, available at NEJM.org). Half the patients were also randomly assigned to receive bevacizumab (15 mg per kilogram of body weight, administered intravenously, every 3 weeks) with each of the first six cycles of chemotherapy and for 10 additional doses every 3 weeks postoperatively (Fig. S1 in the Supplementary Appendix
). Bevacizumab was stopped after cycle 6 for a washout before surgery in order to reduce the risk of surgical complications.
Patients who were considered to be candidates for breast-conserving surgery were to have the primary tumor site marked (with clips or tattoos) before the initiation of chemotherapy. Patients underwent surgery after they had recovered from chemotherapy, the final tumor assessments had been performed, and cardiac function had been evaluated. For patients receiving bevacizumab who underwent all four cycles of doxorubicin– cyclophosphamide, surgery was performed at least 9 weeks after the last dose of bevacizumab. If chemotherapy was stopped before completion of the planned therapy, surgery was performed at least 4 weeks, and preferably 6 weeks, after the last dose of bevacizumab. The type of surgery that was performed was left to the discretion of the patient and surgeon. For patients undergoing breast reconstruction, tissue expansion could not be performed within the 2 weeks before the first postoperative dose of bevacizumab. Expansion or any surgical procedure (e.g., exchanging tissue expanders for permanent implants) was prohibited throughout the course of bevacizumab therapy and a minimum period of 6 weeks after the last dose of bevacizumab. Post-lumpectomy breast radiation therapy was required after breast-conserving surgery. Partial breast irradiation was not permitted. Decisions regarding regional-node irradiation and post-mastectomy irradiation were made at the discretion of the patient's physician. Patients with estrogen-receptor–positive or progesterone-receptor–positive tumors received endocrine therapy for a minimum of 5 years after the completion of neoadjuvant chemotherapy and surgery. The selection of endocrine agents was left to the discretion of the physicians.
The primary end point was the rate of pathological complete response in the breast. Secondary end points included the rate of pathological complete response in the breast and nodes; clinical complete responses after the docetaxel-based portion of the chemotherapy program had been completed; clinical complete responses after completion of the entire sequential chemotherapy program; the percentage of patients with cardiac events, defined as New York Heart Association (NYHA) class III or IV congestive heart failure; and toxic effects, including cardiac events other than congestive heart failure.
Assessments of Tumor Response and Adverse Events
The assessment of tumor response was based on modifications of the criteria proposed by the Response Evaluation Criteria in Solid Tumors Group.17
A pathological complete response in the breast was defined as the absence of histologic evidence of invasive tumor cells in the surgical breast spec imen. A pathological complete response in the breast and nodes was defined as the absence of histologic evidence of invasive tumor cells in the surgical breast specimen, axillary nodes, and nonaxillary sentinel nodes identified after neo-adjuvant chemotherapy. Disease progression was defined as the unequivocal progression of existing target or nontarget lesions; the appearance of one or more new lesions in the breast, regional lymph nodes, or distant sites; or the appearance of inflammatory carcinoma on clinical examination. Adverse events were graded according to the NCI Common Terminology Criteria for Adverse Events, version 3.0 (http://ctep.cancer.gov/protocolDevelopment/electronic_applications/docs/ctcaev3.pdf
There were two primary hypotheses: that the addition of capecitabine or gemcitabine would improve the rate of pathological complete response in the breast, and that the addition of bevacizumab would improve the rate of pathological complete response in the breast. The estimation of the sample size was based on the comparison among chemotherapy regimens. Assuming that the rate of pathological complete response in the docetaxel group would be 26%, we estimated that we would need to enroll 400 patients in each of the three docetaxel-based groups for the study to have 80% power to detect a significantly different rate of 36% for pathological complete response in either the capecitabine group or the gemcitabine group, with a type I error rate of 0.05.
The analyses of end-point data are based on information gathered as of June 30, 2011. The maximum of two standardized pairwise differences in the rates of pathological complete response between the docetaxel group and the other two groups, with or without bevacizumab, was used for testing the improvement in the outcome with the addition of capecitabine or gemcitabine.18
The critical value for a 0.05 significance level is 2.21, which was calculated from 10,000 simulations with adjustment for multiple comparisons.18
The Pearson chi-square test with continuity adjustment19
was used to assess the association between treatment and response variables. The Breslow–Day test was performed to assess the homogeneity of the odds ratios across randomization strata and histologic grades.19
If there was no evidence against the homogeneity of odds ratios, the Mantel–Haenszel estimate of the common odds ratio was calculated in addition to the gross odds ratio.20
All statistical analyses were performed with the use of SAS software, version 9.2, and the R statistical package, version 2.11.