Trial Design and Oversight
We conducted a multicenter, randomized, open-label, phase 3 trial. Eligible patients were randomly assigned to receive T-DM1 or trastuzumab as adjuvant therapy (Fig. S1 in the Supplementary Appendix, available with the full text of this article at NEJM.org).
The trial was designed by a steering committee comprising members of the German Breast Group, the National Surgical Adjuvant Breast and Bowel Project (NSABP) Foundation, independent investigators, and the sponsor (F. Hoffmann–La Roche/Genentech) and was conducted under the guidance of an independent data and safety monitoring committee. Investigators at the trial sites entered data into a database that was held and managed by the NSABP Foundation. The prespecified interim analysis was conducted under the auspices of the data and safety monitoring committee, which recommended full analysis and disclosure of the results. The results and recommendation were reviewed and accepted jointly by the sponsor and members of the trial steering committee, who vouch for the completeness and accuracy of the data and analyses and for the fidelity of the trial to the protocol (available at NEJM.org). The first draft of the manuscript was written by the first author and the last author with assistance from a medical writer paid by the sponsor. All the authors contributed to subsequent drafts. The trial was conducted in accordance with the amended Declaration of Helsinki, and the protocol was approved by the institutional review board at each participating center. All the patients provided written informed consent.
Patients were eligible for participation in the trial if they had histologically confirmed, HER2-positive, nonmetastatic, invasive primary breast cancer (clinical tumor stage T1 to T4, nodal stage N0 to N3, and metastasis stage M0 excluding clinical stage T1aN0 or T1bN0) at presentation and if residual invasive disease was detected pathologically in the surgical specimen of the breast or axillary lymph nodes after completion of taxane-based neoadjuvant chemotherapy administered with trastuzumab. HER2 status was assessed in pretreatment biopsy samples if they were available; if they were not available, a surgical sample was used for assessment. HER2 status was centrally confirmed before trial enrollment (details are provided in the Supplementary Methods section of the Supplementary Appendix). Patients had to have completed at least six cycles (16 weeks) of a conventional preoperative chemotherapy regimen containing a minimum of 9 weeks of taxane-based therapy and 9 weeks of trastuzumab therapy (slightly shorter treatment durations were permitted for dose-dense regimens). Anthracyclines and alkylating agents were permitted according to local standards, as were additional HER2-targeted agents. (Additional eligibility criteria are provided in the Supplementary Methods section of the Supplementary Appendix.) Exclusion criteria included the following: gross residual disease remaining after mastectomy or positive margins after breast-conserving surgery; progressive disease during neoadjuvant therapy; and cardiopulmonary dysfunction, including heart failure of New York Heart Association (NYHA) class II (mild symptoms and function limitation) or higher or a history of a reduction in the left ventricular ejection fraction to less than 40% with previous therapy.
Randomization and Treatment
Within 12 weeks after surgery, patients were randomly assigned in a 1:1 ratio with the use of an interactive voice-response or Web-response system. A permuted-block randomization scheme was used with stratification according to the following: clinical stage at presentation (inoperable breast cancer [tumor stage T4 or nodal stage N2 or N3 and metastasis stage M0] vs. operable breast cancer [tumor stage T1 to T3, nodal stage N0 or N1, and metastasis stage M0]); hormone-receptor status according to local laboratory assessment (estrogen-receptor–positive, progesterone-receptor–positive, or both vs. estrogen-receptor–negative and progesterone-receptor–negative or unknown); preoperative HER2-directed therapy (trastuzumab alone vs. trastuzumab plus an additional HER2-directed agent); and pathological nodal status evaluated after neoadjuvant therapy (node-positive vs. node-negative or not evaluated).
Patients received T-DM1 at a dose of 3.6 mg per kilogram of body weight or trastuzumab at a dose of 6 mg per kilogram intravenously every 3 weeks for 14 cycles. A loading dose of 8 mg of trastuzumab per kilogram was administered if more than 6 weeks had elapsed since the preceding dose of trastuzumab. Patients who discontinued T-DM1 early because of toxic effects could complete 14 cycles of trial treatment with trastuzumab at the discretion of the investigator. Radiation therapy (see the Supplementary Appendix) and endocrine therapy were administered according to institutional standards and the trial protocol.
Imaging assessment for metastatic disease was not mandatory after patients received neoadjuvant therapy and underwent surgery before randomization. Patients were assessed for toxic effects before each dose of trial therapy was administered and during scheduled follow-up visits. The left ventricular ejection fraction was assessed during the last week of cycle 2; every four cycles thereafter; at trial-drug completion (if an assessment was not performed in the previous 6 weeks); at 3, 6, 12, 18, and 24 months; and annually thereafter to year 5. Clinical assessments for disease recurrence occurred every 3 months from the date of randomization to year 2, then every 6 months to year 5, and annually thereafter to year 10.
The primary end point, invasive disease–free survival, was defined as the time from randomization until the date of the first occurrence of one of the following events (hereafter referred to as invasive-disease events): recurrence of ipsilateral invasive breast tumor, recurrence of ipsilateral locoregional invasive breast cancer, contralateral invasive breast cancer, a distant disease recurrence, or death from any cause. The standardized definitions for efficacy end points (STEEP criteria) also include second primary nonbreast cancer as an invasive-disease event,16 so this larger definition of invasive-disease events plus second primary nonbreast cancer was used as a secondary end point. Other secondary end points included disease-free survival (including noninvasive breast cancers), overall survival, distant recurrence–free survival, and safety.
A sample of 1484 patients was planned on the basis of a requirement of 384 invasive-disease events to provide 80% power to detect a hazard ratio of 0.75 with a two-sided significance level of 5% for the primary analysis. This calculation assumed that the percentages of patients who would be free of invasive disease at 3 years would be 70.0% with trastuzumab and 76.5% with T-DM1. A single interim analysis of invasive disease–free survival was planned when approximately 67% of the projected invasive-disease events had occurred, with a P value of 0.0124 for an efficacy stopping boundary or an observed hazard ratio of less than 0.732. The results of the interim analysis crossed the early stopping boundary for benefit of T-DM1 and are presented here.
The primary analysis was based on the intention-to-treat population. An unstratified log-rank test was used to compare invasive disease–free survival between the two treatment groups, as prespecified in the statistical analysis plan, because the smallest subgroup had fewer than five patients in either group. A Cox proportional-hazards model was used to estimate the hazard ratio and its 95% confidence interval. The percentage of patients who would be free of invasive disease at 3 years in each treatment group was estimated with the Kaplan–Meier method. Data from patients who did not have a documented event were censored at the date the patient was last known to be alive and event-free.
The first interim analysis of overall survival was planned to occur if the interim analysis of invasive disease–free survival crossed the prespecified boundary. The overall type I error is controlled at 0.05 for the overall survival analysis with the use of the Lan−DeMets alpha-spending function with an O’Brien−Fleming boundary. Three additional overall survival analyses are planned: an interim analysis at the time of the final invasive disease–free survival analysis (when approximately 384 events have occurred), an interim analysis when approximately 279 deaths have occurred, and a final analysis when approximately 367 deaths have occurred.
The safety analysis included all the patients who received at least one dose of a trial drug. Cardiac events and potential cases of hepatic dysfunction were adjudicated by an independent clinical-events committee. Cardiac events were defined as death from a cardiac cause or heart failure of NYHA class III or IV, with a decrease in the left ventricular ejection fraction of at least 10 percentage points from baseline to a value of less than 50%. These events were summarized according to treatment group.